I need these for studying purposes. @0:16 - Intro @0:40 - What is Dynamic Range? @3:03 - Psychophysics @8:13 - Bending the Curve @12:28 - Gamma and Dynamic Range @16:57 - Let's Talk Log @19:55 - Dynamic Range and ISO @27:10 - Raw and ETTR @30:10 - Demonstration @33:06 - Closing statement
Has a lot, lot , lot of sense. Some people translate the film recording knowledge to the digital world without taking into account the difference in how the light is processed in both medium.
I'm a bit lost at that step. Why not just let more light into the camera and use the lower ISO setting? In the example you say that the ISO 800 image has 6 stop above and 9 below, while the ISO 200 has 4 ab. 11 bel. But why not just use 200 ISO and let 2 more stop of light on the sensor to again achieve 6 stop avobe 9 below?. Is that a consequence of the Log curve? Because definitely what you suggest in the video wouldn't work on a normal DSLR raw.
Whoever those 8 people are that don't like your video, they might just be annoyed that there is such a brilliant resource to get information from. You sir, seriously are the most straight forward and overall correct filmmaking channel not only on youtube but on the entire internet, it seems to me. (Even as an aspiring physicist, i can tell that there wasn't even information that you might find to be incorrect even after a B.Sc. in physics.) Really, I'm impressed.
Exceptionally great video! I can only imagine how much time it took to prepare this video and break this whole behemoth of a subject down into smaller digestible pieces. I think you didn't go the extra mile, you walked an extra marathon! Very informative, yet understandable and easy to watch although being a very complicated topic. I learned so much and it... just felt easy. Thank you so much for taking the time and producing this awesome video!
It's so enjoyable to watch a clip without someone chopping their videos every three words because they just can't talk fluently in front of a camera. I just love watching your clips. I've been following you for many years. This video was a revelation.
I don't know much about filmmaking, but watch these videos with great interest. I *do* know quite a bit about math, electronics, physics and computer science. And always find the quality of these videos impressive and the presentation most enjoyable. To a point of recommending the channel to anybody with a curiosity. Film-buff or not. Even my neighbour who is a senior political writer is now watching it. Thanx!
Easily the best explanation of light/metering/curves/log/raw... you name it. This is gold. Even the "old" way of thinking ISO as one corner of "exposure triangle", replaced with true scientific knowledge... just priceless! I sincerely thank you, Sir, for making this video. Once more, easily the best tutorial if someone wants to truly understand light in cameras, both still and video.
My brain is glowing from the knowledge I just got. Thank you for explaining it in the "big picture" format without making tons of 10 minutes videos to place ads in
I've been shooting since 2015 and i've been mostly running on tips and intuition. Now I've taken a 3 year hiatus from shooting video and photos and I've started to focus on actual science behind process. This video has managed to open my eyes to entirely new possibilities I wasn't aware of! This is the best technical video I've ever come across on TH-cam. You should be absolutely proud of yourself. Now I finally understand that I've been using max 30% of my cameras capabilities instead of the 95% I thought I did. I thank you from the bottom of my heart.
I am subscribed to well over 500 channels (I've lost count), and as such I find myself eligible to say the following: This channel and "engineerguy" are the finest channels on TH-cam. They're so educative and professional I find them like something I would pay 300 bucks to be tutored. Truly an amazing experience watching you guys, you're making the whole YT community better. When I say finest I know some of you are going to say... well there's this one and there's that one... yes, there is, but they got the publicity they wanted, and it pays off, but these two don't have nearly as much as they deserve, and every ONE of you needs to respect that. Their target audience is narrower, aimed towards more professional individuals, they really go the distance when it comes to explaining. Nothing but respect... no but's.
Every 6-months or so I come back to watch these videos to refresh my understanding of exposure, gain and dynamic range. And each time I watch there is a new "Ah Ha" moment when I recall a shooting situation where I gapped on an important part of this. I typically shoot RAW/ETTR but for quick turnaround jobs I will switch to camera processing and then "forget" higher ISO for brighter scenes and lower ISO for darker scenes for example. Thank you for taking the time to explain this, particularly to an old film guys like me who is trying to unlearn the exposure triangle myth.
My perception of this video is also exponential: When you add 1 unit of information my brain needs to process double the amount of information :) Nah, just kidding, this was actually one of the most informative videos in TH-cam history. I could listen to you for 8 hours straight.
Mind blown as it throws out everything I teach regarding ISO. Never been more excited to test a hypothesis, teach on John.. teach on... thank you for taking the time to put this together.
i've learned that from my c200 lol. Oddly i was indoors and shot at iso 4000, the room was well lit. And i didn't see any noise. I thought the camera was good at high iso, i went outside in the dark and shot at iso 4000. It was noise everywhere. I learned that iso dont add noise but amplifies it. Just like how John stated in this video. if your mid tones are at that certain ire where the noise floor is at, which for the c200 is about 12 to 15, then you will have noise especially at high iso. It's a trial and error process hahaha.
but what are you doing if your outdoors, aperture wide open, shutter speed adjusted etc and you are still underexposed? up the gain in post or in camera?
EVERYTHING in filmmaking is a compromise... There's no way around it. If your outdoors and need more light you could drag out some generators and some hmis to add more light or push ISO knowing you're going to take on more noise in your shadows.
I’ve always known about protecting shadows with lower ISO just from experience of boosting exposure in post. Now that I think about it, I have also experienced how you can protect highlights with higher ISO but that just never sounded right so I never paid attention to that possibility. I did notice how you can shoot bright scenes with a high ISO anyway, and the blown out parts are actually easier to recover than the blown out parts from a low ISO image. Now I understand you’re just shifting the amount of data dedicated to highlight versus shadows. You might as well change your channel name to film making myth buster at this point.
@@FilmmakerIQ The new BMPCC 4k camera manual has a graph where it shows how many stops it offers for highlights and how many for shadows in each ISO step. Given that graph, it's best to shoot at ISO 1000 or ISO 1250 with that camera to get the best highlights (and with a variable ND filter, of course, to control the light).
@@FilmmakerIQ I don't understand how increase ISO doesn't reduce DR. By increasing ISO from 200 to 400, camera has half of information to work with. Like you said in the video it raise up the noise floor thus reduce DR by one stop. You also then mention there are other factors that contribute noise as well, so it seems the penalty of one stop DR reduction is the best case scenario. However you then said increasing ISO doesn't reduce DR. That makes me go "huh?" The way I understand is that the sensor can capture much greater DR(let's say 10 stops) than the log or gamma profile(8 stops) can provide. By raising the ISO you simply change where the grey point is, increase noise floor and because it hasn't hit the max signal range yet, there is no penalty of highlight. Basically if you shoot log, there is no penalty in DR if you raise from ISO 200 to 800, but you will if you raise to 1600. It also means if you shoot Raw, then you might as well just stay in base ISO, get the full DR, preserve high light and since most space is for shadow details anything, just raise up in post. Thank you in advance for reading and answering my question. :)
Some people probably think this is too geeky, but it's actually super awesome and fascinating!! EDIT: Also feel free to correct me for what I think I understood: 1) ETTR is about increasing Signal to Noise ratio (hence cleaner images) by increasing exposure using only shutter speed and/or aperture, then bringing down exposure in post. 2) Higher ISO gives more details (i.e. better dynamic range) in the highlights, whilst Lower ISO gives more detail (i.e. better dynamic range) in the shadows.
Even a term like "native ISO" is sort of confusing... There's really no such thing - it breaks down to what ISO gives you the even number of stops above and below middle gray. Then Native ISO is also dependent on what flavor of LOG you're shooting at... C-Log it's 800 or so, S-Log it's like 2500 (I think it locks you into that)! It's all confusing!! :P :P
Agreed. I guess we should really say optimal ISO. S-Log 2 locks you into 800, while S-Log 3 locks you into 1600. I did some experimenting with a neutral gray card and light meter. The exposure with the lowest noise floor and highest dynamic range for S-Log 2 was 2 stops over exposed S-Log 3 was 3 stops over, and S-Log 3 has one extra stop of dynamic range and is cleaner when properly expose, which makes sense.
Watched the whole video. Insanely informative. It's nice to not only know the how, but also the why. And those tips for shooting in Log were KEY! Thank you. 🙏🏻
Dude, I'm forever grateful for this youtube channel. The amount of knowledge put out in such high quality content FOR FREE is mind blowing! I love you and all your patrons!
Thank you so much for this! The "Red Pill Moment" described below at your kicker at 26:19 has been making me think of the "Zone" method of calculating exposure and dynamic range. Most popularized by Ansel Adams, using individual 8x10 inch photographic plates per image, the photographer used a spot meter to calculate the dynamic range required of the negative to allocate as much detail as the scene required, and would then 'rate' the ISO of the plate accordingly - providing the printer with all the juicy dynamic range needed for producing an incredibly rich final image. I'm no doubt wrong on several fronts here, but it just sounds a bit familiar with is explained here...
this is the most in-depth yet understandable explanation of Digital ISO/RAW on TH-cam! Something that isn't widely discussed: Why is ISO unchangeable in still photo RAW processing applications yet ISO is an adjustable piece of metadata based on native sensor ISO while working with cinema camera RAW files like BRAW and R3D in non-linear editors?
That's a really good question and I think the answer is really going to be one of semantics (because you are effectively adjusting the ISO in a Raw editor when you play with the exposure slider). Digital moving pictures have a background in video where people think in terms of a signal where as still photography still thinks in terms of a frame of celluloid. Gaining up and down the recorded signal (which some software engineer thought would be useful to translate that into ISO) naturally makes sense but maybe not so in stills even though the actual engineering is exactly the same!
You're officially my hero, watched some of your newer videos on audio and youTube etc. was really impressed. This one, I may have watched a couple of years ago, but fully appreciate it now ( my learning curve has been raised.) thank you. I may watch this periodically to recall the details later.
Started watching your courses during the lockdown. They really are excellent. Everyone in Production and Post should watch these ! Thanks for creating.
All this hard work for these videos...to help every one of us who love photography and filmmaking to understand the physics and the whole theory behind the art! I want to say something more than THANK YOU but I cannot find the words!
John, I can't believe you've done this. This was so informative and very well explained. You blew my mind with the higher/lower iso and dynamic range part. I definitely have to watch this video a few times, so I understand it well enough to explain to others. Thank you.
Thank you for taking the time, spending the energy and brain power to put all of this together in a coherent video! This is EXCELLNT work! This is the side of TH-cam I enjoy! Keep up the GREAT work! 👍
Sometime back I took issue with a single assertion you made about the visibility of jitter. (I still do.) Other than that, I found the video to be excellent. I was prompted to watch more of your videos. They are universally EXCELLENT! I don't know anything about films you've been involved in making but I suspect they're great as each of the "Filmmaker IQ" videos I've watched has been informative, well researched, well paced and a delight to watch. FIVE GOLD STARS! 🌟🌟🌟🌟🌟
The phenomenon you describe starting at @24:00 of highlight headroom progressively increasing as ISO increases only applies on camera systems that use digital scaling exclusively for their ISO implementations. There are only a few cameras I've found which do this, including those from Blackmagic. Most cameras, including those from Canon, Nikon, Sony, and Panasonic use digital scaling at their native log ISOs, switching to a combination of digital scaling + analog amplification thereafter for all ISOs above native. For example, Nikon's N-Log has a native ISO of 800, which corresponds to sensor ISO 100 (in terms of gain). This means N-Log ISO 800 is ISO 100 with 3EV of underexposure when you account for the metering system using notational ISO 800. The brightness is adjusted +3EV via digital scaling after reading the frame from the sensor, to achieve the ISO 800 exposure rating - this is done digitally instead of via analog gain so that the highlights aren't clipped on the bit-depth limited ADC prior to encoding them via the log curve. Switching topics, the highlight headroom phenomenon you demonstrated on you Canon for ISOs below its native C-Log of ISO 800 is unrelated to the scaling I just described. The Canon reduction in highlights below its native Clog ISO 800 is due to digital pulling, the same as how the expanded ISOs are implemented on Canon cameras for still images. For example, at ISO 200 the Canon meters for notational ISO 200, which induces 2EV of clipped highlights relative to the native Clog ISO of 800. This is also why you observe a hard-stop of IRE levels for those expanded ISOs. At 26:18 you compare Canon's ISO 100 to ISO 1600 to demonstrtate expanded highlight headroom from the higher ISO, but that highlight headroom expansion is only due to the highlight clipping from ISO 100 to 800. If you compare your Canon's ISO 800 to any other higher ISO you will not observe any increase in highlight headroom. This is consistent with the Canon DR chart you show on-screen @ 26:10. I'm working on a video that does a deep technical dive into how cameras implement log internally as an adjunct to your fantastic video. I'd love to discuss a few technical details with you if you're able to find the time. You can contact me via PM on Dpreview - my username there is "Horshack".
It's primary in the line of cameras advertised as cinema camera, you'll see in Alexa lines, Black magic, Red. The Canon C line demonstrates it as you noted in sub 800 ISOs... But it is an key caveat to point this out there because outside on a bright sunny day is where you would most likely reach for ISO100 where you will get the most clipping of highlights.
@@FilmmakerIQ Thanks John. IMO shooters should avoid the non-native log ISOs. They provide no advantage vs the native ISO w/digital exposure adjustment in post. Also, since the log encoding is clipped, all the highlight-related exposure aid thresholds must be adjusted every time the the camera ISO is switched in or out of the non-native ISOs. For example, having to adjust the configured threshold value for highlight zebras, or adjusting which color represents highlight saturation on the false-color display, or having to visually adjust how their interpret the waveform to account for the lower maximum IRE.
Well it depends... If you're shooting on cameras that behave the way I described that push and pull, there is no disadvantage to shooting non native ISO. You have to understand your own camera and what it's doing
Umm John, I am re-watching this video because your explanation of raising the ISO was over my head. If I am not mistaken, or getting crazy, you updated it and added new elements, especially the cloud example. Now I get it. Thanks so much!. This is a great video!. Cheers
@@FilmmakerIQ "Unfortunately can't add or edit TH-cam Videos... else I'd fix a bunch of stuff lol" 🙂 Change "TH-cam Videos" to "past actions" and you have my number one frustration with life. th-cam.com/users/sgaming/emoji/7ff574f2/emoji_u1f642.png
Thanks for the informative (and counter-intuitive!) video, and even bigger thanks for discussing the ISO issue in the comments so patiently. The latter really helped me start to get my head around it.
MIND BLOWING...!!! Lower your exposure, higher ISO to protect highlights, Open up your exposure lower ISO to protect your shadows,.....Am Jumping up and down joyfully for just learning this GEM....Thanks
This is one awesome video. I cannot thank you enough; this information was extremely well presented and the presenter was great, too. Thank you so much!!!!
Your videos are SO GOOD! SO SO GOOD! =) Total pleasure to watch ads on your video. Totally worth and hope you can continue to make videos like this. Sharing knowledge is what will drive us forward.
ThankYou soo much for this! Watched your previous video and this one and loved both! Your style, teaching method, editing, scientific research is perfect! Look forward to more and like everyone else said the higher iso to protect the highlights and lower for the shadows is blowing my mind
Wow, you are a great teacher. I went from knowing "a lot" about photography, but being baffled (it turns out) by the subjects you covered, to having a basic understanding about the subjects you covered. Thank you.
Great Video and Finally someone who underlines what I said ever since I red an Article on Red's website concerning ISO. If you want to preserve highlight use a higher ISO and if you want to preserve shadows use a lower ISO. I felt Like a madman trying to explain ppl, how the sensors work, but now I will just link to your video instead! Thank you very much! And I am looking forward to the HDR Video! Cause that's another topic that noone seems to explain properly. I hop you can explain it!
This is a wonderful lesson, well worth the wait since our last one. However, there is just one single thing I'd like to point out regarding BT.709. The BT.709 specifies a standard that includes a specific color gamut (identical to sRGB), white point (D65) and color space (YCbCr), which then also includes the levels of course (SMPTE Legal Levels). However, it have never really had a specified standard for EOTF though 2.2 have certainly been widely used in this regard. The closest we would get to an EOTF standart for BT.709 would be BT.1886, which was passed in 2011. I would also like to point out that certain CRTs have very deviant EOTF in comparison to 2.35-2.55, and they also tend to have very variable EOTF through the greyscale. All of this is to be expected of course as the CRT itself is naturally imprecise in this regard. For instance, the Sony BVM-D24E1WU can go almost as low as 3.0 in the shadow detail as its curve gradually goes towards 2.5 at the brightest highlights (though most of the middle grey is certainly at around 2.7-2.8). Funny enough, the most accurate CRT for I've seen in this regard (with least deviant EOTF curve) is the Sony HDM-3830, which was part of Sonys first HD CRT. It follows Power Law 2.4 almost perfectly, something I've never seen on any other professional grade CRT. Some Ikegamis I've seen have been very close however.
Thanks so much for doing this! Again one of the best learning experience I have ever had ! Have you ever considered making an episode about color space workflow using ACES?
Recién empiezo a estudiar algo de fotografía y no entendía porque muchas paginas me decían que subiera el ISO en situaciones de poca luz cuando por experiencia en post ya me había dado cuenta que era obtenía mas información haciendo lo contrario. Gran explicación!
Awesome work, I think I'll have physics students watch this video, just taught sound intensity and level (dB) and this would be a good follow on topic!
The advice @26:15 (shoot at higher ISO to preserve highlights) is going to sound counterintuitive to vloggers who are used to shooting ETTR with wide open apertures (to minimize depth of field). Most shooters are familiar with sensor dynamic range charts that show maximum DR occurs at the lowest native ISO settings. If you follow those charts, you'll set ISO at minimum, shutter at 180-degrees, open the aperture as wide as you can, and use an ND filter to prevent the highlights from clipping. That will give you ETTR without reference to middle gray level, which vloggers typically ignore. The workflow recommended in this video is based on the assumption that you are following the cinematography convention of exposing middle gray at around 38%. The rationale for that practice is not necessarily to maximize DR for each shot, but to establish a common middle gray reference point to make intercuts between footage consistent. With that workflow, raising the ISO will increase the brightness of middle gray, and you will have to lower your exposure (by narrowing the aperture or using more ND) in order to push middle gray down to 38%. It is that adjustment that gives you more highlight range above middle gray. What actually produces more highlight headroom is not simply raising the ISO, but lowering the amount of light that is passing through the lens to the sensor.
This is incorrect - it's fundamentally flawed way of looking at the light. The problem is ND filters and exposures don't attack just highlight range - the apply throughout the range evenly. If you reduce the light to accomodate the highlights, you reduce the light in the mid tones and shadows at the same time. Yes, raising the ISO will increase the brightness of middle gray. But reducing the exposure to counteract a raise in ISO doesn't actually protect your highlights - it does nothing. Visualize it this way: You have a middle gray and a highlight +5 stops above. You raise your ISO, so now what was middle gray is +1 stop above... now your highlight is +6 stops... so you reduce your exposure... you're back to middle gray with a highlight at +5 stops. Thus the only thing you did was raise the noise floor by 1 stop if you did that... This is with the traditional understanding of ISO, but many modern cameras (especially "cinema cameras") utilize a push pull system for ISO where the dynamic range is consistent throughout a chunk of ISO range. This article using my video on the BMPCC 4K is an example of this: www.premiumbeat.com/blog/dual-native-iso-explained/ You'll find similar charts on Red/Arri/Canon cameras as well. Those cameras will allocate more of the dynamic range shades above middle gray as you step up the ISO. I also explain it all at 24:12 in this video which you should revisit. Finally middle gray isn't a cinematography convention - it's a photography convention that goes back to the origins. And middle gray is central to the concept of defining ISO as well. Film timing to correct and get consistent exposures has been a thing since forever - it's not why we shoot for middle gray. I personally would advocate to put the preference to middle gray and then just look out for highlights and shadows that fall out of range and try to wrangle them in. You can try to expose in such a way to capture every bit of dynamic range but I think that misses the forest for the trees.
@@FilmmakerIQ - "But reducing the exposure to counteract a raise in ISO doesn't actually protect your highlights - it does nothing." Ah, but it does if you reduce exposure by actually lowering the amount of light that passes through the lens (i.e. by narrowing the aperture or increasing the ND filter). That is how you get an extra stop of highlight headroom, by making room for more stops of light intensity above middle gray, not simply by increasing ISO. Notice, I'm not talking about increasing DR here, only about increasing highlight headroom above middle gray, which as you advised, is the point of raising the ISO. What is actually "non-traditional" about BlackMagic's chart of ISO vs dynamic range is their virtualization of the concept of ISO. The reason the dynamic range of the BMPCC stays constant across a range of ISO's in their chart is because they're not actually recording at incremental analog and/or digital gain levels, they're recording at the native ISO for each range, and using metadata to record the virtual ISO for decoding purposes. In other words, the same range of light intensities (i.e. what is actually detected by the sensor) is being recorded at each ISO step in the range, the only thing that changes is the designated level of "middle gray". But since middle gray is the reference point upon which all other DR measurements are based, everything else changes accordingly.
_but it does if you reduce exposure by actually lowering the amount of light that passes through the lens. That is how you get an extra stop of highlight headroom_ NO IT DOESN'T. You are not giving yourself headroom by changing the exposure, that's thinking by putting the cart before the horse. The light coming in from the lens will have the same dynamic range regardless of ND filters and exposure settings. Proportions between tone ranges are not changing. It's the ISO that determines the headroom. Adjusting the sensor setting gives you allocation of the highlights and shadows, not the exposure. You match the exposure for the ISO but the exposure is NOT responsible for giving you headroom (unless you purposely underexpose). In other words - if you just increase the ISO (on these cameras that push/pull for ISO) without changing the exposure - you will increase the headroom above middle gray... but everything will be just overexposed. The headroom above middle gray exists whether you utilize it or not. I think that is where you confusion is - there is no "middle gray" in terms of light from the scene - it's only in relation to how we set up the sensor. Vloggers would do best to really understand what I'm saying - what you're saying is just adding to confusion.
@@FilmmakerIQ "The light coming in from the lens will have the same dynamic range regardless of ND filters and exposure settings. Proportions between tone ranges are not changing." It's not the dynamic range of the light itself, it's the maximum luminance the sensor can detect before it saturates (i.e. the limits of its sensitivity to light), divided by its noise floor, that determines the native dynamic range of the camera. As you point out, a Dual Native ISO camera has two separate sensitivity ranges, but within each range its native sensor sensitivity cannot be changed by ISO settings. Once you reach the maximum luminance that sensor range can handle, the ONLY way you can increase the maximum light level it can discriminate is to lower the amount of light passed through the lens (i.e. by narrowing the lens aperture or increasing the ND filter). While it's correct to say "it's the ISO that determines the headroom", the BMPCC Dynamic Range vs ISO chart illustrates what the camera's ISO setting is actually determining. Within each sensitivity range, the camera's dynamic range remains unchanged. The ONLY thing that changes when you change the ISO is the designated level assigned to middle gray within each range (until you get to extremely high ISO's). My hat's off to BlackMagic for making it unmistakably clear what the camera's ISO setting actually does.
I think I see where your argument is going but where this is getting flummoxed is you're trying to avoid middle gray. _Once you reach the maximum luminance that sensor range can handle, the ONLY way you can increase the maximum light level it can discriminate is to lower the amount of light passed through the lens_ Agreed. But that says nothing about headroom. Headroom is defined by the range above middle gray. What is middle gray is ultimately is the question. Every ISO has a maximum light input... On the BMPCC example ISO 100 an ISO 1000 both have the same maximum light level it can discriminate - ISO 1000 has a lot more headroom between middle gray and full saturation than ISO 100... But if you were to completely ignore middle gray and worry only about staying just below the clipping point (ETTR)- the idea that reducing the light that lands on the sensor so that's it's further from the clipping point works out as way of "protecting your highlights" Taken to the extreme, it doesn't matter what ISO you shoot on the BMPCC, so long as you anchor your exposure to just under the clipping point of which ever "native ISO" you chose. The resulting images will be underexposed light scenes and over exposed dark scenes - which you can in theory fix in post. But I think that approach is fraught with dangers and creates more headaches then it solves actually trying to expose for middle gray. But honestly, I don't know where this conversation is heading or what you were trying to correct me on. ISO determines dynamic range, you expose for the ISO - I haven't seen anything to counter that line of thinking :)
@7:31 I'm really struggling to understand how you get the orange curve. I get lost when you say "Measure the distance between each stop of light". What am I missing? Can you help explain this for me? Wonderful video and thank you for putting it together.
I have two questions which i am slighlty confused about. 1. In stevesn power law there is k which represents the constant of proportionality, but hwat does that actually represent and what number did u use and why when calculating the perceived light sensation. 2. at 7;40- i get how you achieved the first graph by plotting the perceived light sensation on the x ais and teh light intesnity of different sorces on the y axis but what did u mean by teh change in power curve at 7:55-
1. K is just a scaling constant and it changes depending on the units used. For this discussion, all that matters is it's a positive number. 2. Plotting the "change" in the second graph is asking, how much do we "lose in experience" as we step down. From 100% to 50% (1 stop loss), we only lose an experience of "0.2" from 1.0 to 0.8 on the y axis) From 50% to 25% we lose about "0.16" on the y graph - then "1.2" for the next stop. The point is even though we are reducing the light exponentially, the experience stays relatively linearly constant - it's not flat, but it looks like it evenly steps down in brightness even though in actually light units, it's decreasing geometrically.
Mr Hess, how do I properly express how amazing your videos are and how grateful I am and how grateful I am sure everyone else is? Hmm let me think, you deserve something good. Okay, got it. Here goes: If I were a woman, and if I more specifically were a Jessica Rabbit style femme fatale woman, I'd come over and thank you in the best way, totally rock your world. :D GREAT work!
Took me back to 1984 when I constructed my first video frame grabber. I experienced ALL of the issues you discussed here. It was a good ten years later, as a professional broadcast equipment designer, that I caught up with the theory behind it. However, early on, using only my squinty eyeballs and pure experimentation, I created very effective gamma correction values to apply to the look up tables built into the color D/A converters! Even though, I had only the vaguest notion of all the psycho visual factors involved. I loved every second of it! (Don't even get me started on all the nuance of aspect ratios! Pixels vs clock rates vs screen dimensions, etc! Why aren't my calculated circles actually round? ARRRGGGHH!) Thanks!!!!
Canon has this "Save Highlights" option, which basically shoots at one ISO higher than normal and then reduces the exposure in camera one stop. Makes totally sense now.
As a retired TV engineer who spent 30 years inside an assortment of cameras, this was a great video! Best explanation I have seen. Thanks.
I need these for studying purposes.
@0:16 - Intro
@0:40 - What is Dynamic Range?
@3:03 - Psychophysics
@8:13 - Bending the Curve
@12:28 - Gamma and Dynamic Range
@16:57 - Let's Talk Log
@19:55 - Dynamic Range and ISO
@27:10 - Raw and ETTR
@30:10 - Demonstration
@33:06 - Closing statement
Red pill moment!
Protect highlights with HIGHER ISO
Protect shadows with LOWER ISO!!!
I puzzled over that for probably about 2+ years... It took that long to build the case for it in my mind and solid enough to present it.
Brain just exploded. Should probably clean that up...
I would have never believed anyone without a presentation like the one in this video, truly eye opening
Has a lot, lot , lot of sense. Some people translate the film recording knowledge to the digital world without taking into account the difference in how the light is processed in both medium.
I'm a bit lost at that step. Why not just let more light into the camera and use the lower ISO setting?
In the example you say that the ISO 800 image has 6 stop above and 9 below, while the ISO 200 has 4 ab. 11 bel.
But why not just use 200 ISO and let 2 more stop of light on the sensor to again achieve 6 stop avobe 9 below?.
Is that a consequence of the Log curve? Because definitely what you suggest in the video wouldn't work on a normal DSLR raw.
Damn! Great video John. Went into way more details then I ever thought possible but really explains it well.
That's a huge amount of information brilliantly visualized! You guys inspire me :)
You went full dork with this one. I love it.
You've taught me more about moviemaking and cinema history than any film school could ever teach. God bless you John.
Whoever those 8 people are that don't like your video, they might just be annoyed that there is such a brilliant resource to get information from. You sir, seriously are the most straight forward and overall correct filmmaking channel not only on youtube but on the entire internet, it seems to me. (Even as an aspiring physicist, i can tell that there wasn't even information that you might find to be incorrect even after a B.Sc. in physics.)
Really, I'm impressed.
Exceptionally great video!
I can only imagine how much time it took to prepare this video and break this whole behemoth of a subject down into smaller digestible pieces. I think you didn't go the extra mile, you walked an extra marathon!
Very informative, yet understandable and easy to watch although being a very complicated topic. I learned so much and it... just felt easy.
Thank you so much for taking the time and producing this awesome video!
This could be the best, the most explanatory and easiest-to-grasp video about dynamic range. Thank you so much Mr. Hess, you rock!
It's so enjoyable to watch a clip without someone chopping their videos every three words because they just can't talk fluently in front of a camera. I just love watching your clips. I've been following you for many years. This video was a revelation.
I don't know much about filmmaking, but watch these videos with great interest. I *do* know quite a bit about math, electronics, physics and computer science. And always find the quality of these videos impressive and the presentation most enjoyable. To a point of recommending the channel to anybody with a curiosity. Film-buff or not. Even my neighbour who is a senior political writer is now watching it. Thanx!
Easily the best explanation of light/metering/curves/log/raw... you name it. This is gold. Even the "old" way of thinking ISO as one corner of "exposure triangle", replaced with true scientific knowledge... just priceless! I sincerely thank you, Sir, for making this video. Once more, easily the best tutorial if someone wants to truly understand light in cameras, both still and video.
My brain is glowing from the knowledge I just got. Thank you for explaining it in the "big picture" format without making tons of 10 minutes videos to place ads in
I've been shooting since 2015 and i've been mostly running on tips and intuition. Now I've taken a 3 year hiatus from shooting video and photos and I've started to focus on actual science behind process.
This video has managed to open my eyes to entirely new possibilities I wasn't aware of! This is the best technical video I've ever come across on TH-cam. You should be absolutely proud of yourself.
Now I finally understand that I've been using max 30% of my cameras capabilities instead of the 95% I thought I did. I thank you from the bottom of my heart.
Thank you. The same thing happened to me once I really started to understand how this stuff works.
I am subscribed to well over 500 channels (I've lost count), and as such I find myself eligible to say the following:
This channel and "engineerguy" are the finest channels on TH-cam. They're so educative and professional I find them like something I would pay 300 bucks to be tutored. Truly an amazing experience watching you guys, you're making the whole YT community better.
When I say finest I know some of you are going to say... well there's this one and there's that one... yes, there is, but they got the publicity they wanted, and it pays off, but these two don't have nearly as much as they deserve, and every ONE of you needs to respect that. Their target audience is narrower, aimed towards more professional individuals, they really go the distance when it comes to explaining. Nothing but respect... no but's.
This is one of the best-presented videos I've ever seen on TH-cam. Amazing job! I learned so much, I'll have to watch again.
I cannot even believe this video. I am in AWE! How do you not have 1 million subscribers yet? You are a very smart, well spoken instructor! Wow!
Every 6-months or so I come back to watch these videos to refresh my understanding of exposure, gain and dynamic range. And each time I watch there is a new "Ah Ha" moment when I recall a shooting situation where I gapped on an important part of this. I typically shoot RAW/ETTR but for quick turnaround jobs I will switch to camera processing and then "forget" higher ISO for brighter scenes and lower ISO for darker scenes for example. Thank you for taking the time to explain this, particularly to an old film guys like me who is trying to unlearn the exposure triangle myth.
Greatest filmmaking TH-cam channel
A masterpiece in 34 minutes. Thank you for breaking such much down John.
maybe one of the most helpful videos on ISO and dynamic range. Thank You so much.
This channel deserves millions of subscribers and patreons. Great explanations. Thank you. I learned a lot.
My perception of this video is also exponential: When you add 1 unit of information my brain needs to process double the amount of information :)
Nah, just kidding, this was actually one of the most informative videos in TH-cam history. I could listen to you for 8 hours straight.
Mind blown as it throws out everything I teach regarding ISO. Never been more excited to test a hypothesis, teach on John.. teach on... thank you for taking the time to put this together.
34 Minutes ... Maaaan, you saved my evening. Thank you!
I'm addicted to this channel. superb work!
My life will never be the same: use lower ISO in the dark, and higher ISO in bright cond. *breathing heavily*
i've learned that from my c200 lol. Oddly i was indoors and shot at iso 4000, the room was well lit. And i didn't see any noise. I thought the camera was good at high iso, i went outside in the dark and shot at iso 4000. It was noise everywhere. I learned that iso dont add noise but amplifies it. Just like how John stated in this video. if your mid tones are at that certain ire where the noise floor is at, which for the c200 is about 12 to 15, then you will have noise especially at high iso. It's a trial and error process hahaha.
What's more, it's really the noise floor we object to... Not necessarily the noise in the signal. ;)
but what are you doing if your outdoors, aperture wide open, shutter speed adjusted etc and you are still underexposed? up the gain in post or in camera?
You know how they say - "when you gotta go - you gotta go!". This applies to ISO as well.
EVERYTHING in filmmaking is a compromise... There's no way around it.
If your outdoors and need more light you could drag out some generators and some hmis to add more light or push ISO knowing you're going to take on more noise in your shadows.
I’ve always known about protecting shadows with lower ISO just from experience of boosting exposure in post. Now that I think about it, I have also experienced how you can protect highlights with higher ISO but that just never sounded right so I never paid attention to that possibility. I did notice how you can shoot bright scenes with a high ISO anyway, and the blown out parts are actually easier to recover than the blown out parts from a low ISO image. Now I understand you’re just shifting the amount of data dedicated to highlight versus shadows.
You might as well change your channel name to film making myth buster at this point.
Higher ISO just to the point where you're not tapping down the whites... I mean don't go nuts and shoot ISO 512000 in mid day sun hahaha.
@@FilmmakerIQ The new BMPCC 4k camera manual has a graph where it shows how many stops it offers for highlights and how many for shadows in each ISO step. Given that graph, it's best to shoot at ISO 1000 or ISO 1250 with that camera to get the best highlights (and with a variable ND filter, of course, to control the light).
I have that camera on rental order and going to do a quick video on it
@@FilmmakerIQ I don't understand how increase ISO doesn't reduce DR. By increasing ISO from 200 to 400, camera has half of information to work with. Like you said in the video it raise up the noise floor thus reduce DR by one stop. You also then mention there are other factors that contribute noise as well, so it seems the penalty of one stop DR reduction is the best case scenario. However you then said increasing ISO doesn't reduce DR. That makes me go "huh?"
The way I understand is that the sensor can capture much greater DR(let's say 10 stops) than the log or gamma profile(8 stops) can provide. By raising the ISO you simply change where the grey point is, increase noise floor and because it hasn't hit the max signal range yet, there is no penalty of highlight.
Basically if you shoot log, there is no penalty in DR if you raise from ISO 200 to 800, but you will if you raise to 1600.
It also means if you shoot Raw, then you might as well just stay in base ISO, get the full DR, preserve high light and since most space is for shadow details anything, just raise up in post.
Thank you in advance for reading and answering my question. :)
Yes
Thank you! I believe this is one of the most under-rated TH-cam channels. Should have more subscribers and views.
Some people probably think this is too geeky, but it's actually super awesome and fascinating!!
EDIT: Also feel free to correct me for what I think I understood:
1) ETTR is about increasing Signal to Noise ratio (hence cleaner images) by increasing exposure using only shutter speed and/or aperture, then bringing down exposure in post.
2) Higher ISO gives more details (i.e. better dynamic range) in the highlights, whilst Lower ISO gives more detail (i.e. better dynamic range) in the shadows.
Great video, John! This is why knowing your camera sensor's "native" ISO is so important for reducing noise and maximizing dynamic range.
Even a term like "native ISO" is sort of confusing... There's really no such thing - it breaks down to what ISO gives you the even number of stops above and below middle gray. Then Native ISO is also dependent on what flavor of LOG you're shooting at... C-Log it's 800 or so, S-Log it's like 2500 (I think it locks you into that)!
It's all confusing!! :P :P
Agreed. I guess we should really say optimal ISO. S-Log 2 locks you into 800, while S-Log 3 locks you into 1600. I did some experimenting with a neutral gray card and light meter. The exposure with the lowest noise floor and highest dynamic range for S-Log 2 was 2 stops over exposed S-Log 3 was 3 stops over, and S-Log 3 has one extra stop of dynamic range and is cleaner when properly expose, which makes sense.
Most likely the best free video on this subject.
Watched the whole video. Insanely informative. It's nice to not only know the how, but also the why. And those tips for shooting in Log were KEY! Thank you. 🙏🏻
This is the best explanation of log and DR I've seen - Fantastic job!
Dude, I'm forever grateful for this youtube channel. The amount of knowledge put out in such high quality content FOR FREE is mind blowing! I love you and all your patrons!
Thank you so much for this! The "Red Pill Moment" described below at your kicker at 26:19 has been making me think of the "Zone" method of calculating exposure and dynamic range. Most popularized by Ansel Adams, using individual 8x10 inch photographic plates per image, the photographer used a spot meter to calculate the dynamic range required of the negative to allocate as much detail as the scene required, and would then 'rate' the ISO of the plate accordingly - providing the printer with all the juicy dynamic range needed for producing an incredibly rich final image.
I'm no doubt wrong on several fronts here, but it just sounds a bit familiar with is explained here...
Brilliant video. I've never seen an explanation as clear as this. Lost count of how many times I thought 'so that's how it works!'
Love this channel.
This is food for my brain. Thanks. One of the best videos ever seen about dynamic range.
Ah, finally one of the good old lecture vids again. Thanks for that great explanation.
this is the most in-depth yet understandable explanation of Digital ISO/RAW on TH-cam! Something that isn't widely discussed: Why is ISO unchangeable in still photo RAW processing applications yet ISO is an adjustable piece of metadata based on native sensor ISO while working with cinema camera RAW files like BRAW and R3D in non-linear editors?
That's a really good question and I think the answer is really going to be one of semantics (because you are effectively adjusting the ISO in a Raw editor when you play with the exposure slider). Digital moving pictures have a background in video where people think in terms of a signal where as still photography still thinks in terms of a frame of celluloid. Gaining up and down the recorded signal (which some software engineer thought would be useful to translate that into ISO) naturally makes sense but maybe not so in stills even though the actual engineering is exactly the same!
You're officially my hero, watched some of your newer videos on audio and youTube etc. was really impressed. This one, I may have watched a couple of years ago, but fully appreciate it now ( my learning curve has been raised.) thank you. I may watch this periodically to recall the details later.
Started watching your courses during the lockdown. They really are excellent. Everyone in Production and Post should watch these ! Thanks for creating.
All this hard work for these videos...to help every one of us who love photography and filmmaking to understand the physics and the whole theory behind the art! I want to say something more than THANK YOU but I cannot find the words!
John, I can't believe you've done this.
This was so informative and very well explained.
You blew my mind with the higher/lower iso and dynamic range part.
I definitely have to watch this video a few times, so I understand it well enough to explain to others. Thank you.
Thank you for taking the time, spending the energy and brain power to put all of this together in a coherent video! This is EXCELLNT work! This is the side of TH-cam I enjoy! Keep up the GREAT work! 👍
Thank you so much!
Sometime back I took issue with a single assertion you made about the visibility of jitter. (I still do.) Other than that, I found the video to be excellent. I was prompted to watch more of your videos. They are universally EXCELLENT! I don't know anything about films you've been involved in making but I suspect they're great as each of the "Filmmaker IQ" videos I've watched has been informative, well researched, well paced and a delight to watch.
FIVE GOLD STARS! 🌟🌟🌟🌟🌟
This video was simply brilliant! Really well explained! Thanks John!
Thank god I didn't go to film school. Thanks for this beautiful video! Explains it waaaay better than a lecture could
The phenomenon you describe starting at @24:00 of highlight headroom progressively increasing as ISO increases only applies on camera systems that use digital scaling exclusively for their ISO implementations. There are only a few cameras I've found which do this, including those from Blackmagic. Most cameras, including those from Canon, Nikon, Sony, and Panasonic use digital scaling at their native log ISOs, switching to a combination of digital scaling + analog amplification thereafter for all ISOs above native. For example, Nikon's N-Log has a native ISO of 800, which corresponds to sensor ISO 100 (in terms of gain). This means N-Log ISO 800 is ISO 100 with 3EV of underexposure when you account for the metering system using notational ISO 800. The brightness is adjusted +3EV via digital scaling after reading the frame from the sensor, to achieve the ISO 800 exposure rating - this is done digitally instead of via analog gain so that the highlights aren't clipped on the bit-depth limited ADC prior to encoding them via the log curve. Switching topics, the highlight headroom phenomenon you demonstrated on you Canon for ISOs below its native C-Log of ISO 800 is unrelated to the scaling I just described. The Canon reduction in highlights below its native Clog ISO 800 is due to digital pulling, the same as how the expanded ISOs are implemented on Canon cameras for still images. For example, at ISO 200 the Canon meters for notational ISO 200, which induces 2EV of clipped highlights relative to the native Clog ISO of 800. This is also why you observe a hard-stop of IRE levels for those expanded ISOs. At 26:18 you compare Canon's ISO 100 to ISO 1600 to demonstrtate expanded highlight headroom from the higher ISO, but that highlight headroom expansion is only due to the highlight clipping from ISO 100 to 800. If you compare your Canon's ISO 800 to any other higher ISO you will not observe any increase in highlight headroom. This is consistent with the Canon DR chart you show on-screen @ 26:10. I'm working on a video that does a deep technical dive into how cameras implement log internally as an adjunct to your fantastic video. I'd love to discuss a few technical details with you if you're able to find the time. You can contact me via PM on Dpreview - my username there is "Horshack".
It's primary in the line of cameras advertised as cinema camera, you'll see in Alexa lines, Black magic, Red. The Canon C line demonstrates it as you noted in sub 800 ISOs... But it is an key caveat to point this out there because outside on a bright sunny day is where you would most likely reach for ISO100 where you will get the most clipping of highlights.
@@FilmmakerIQ Thanks John. IMO shooters should avoid the non-native log ISOs. They provide no advantage vs the native ISO w/digital exposure adjustment in post. Also, since the log encoding is clipped, all the highlight-related exposure aid thresholds must be adjusted every time the the camera ISO is switched in or out of the non-native ISOs. For example, having to adjust the configured threshold value for highlight zebras, or adjusting which color represents highlight saturation on the false-color display, or having to visually adjust how their interpret the waveform to account for the lower maximum IRE.
Well it depends... If you're shooting on cameras that behave the way I described that push and pull, there is no disadvantage to shooting non native ISO. You have to understand your own camera and what it's doing
I'd love to see an episode on audio dynamic range, talking about improvements in cinema audio technology over the years.
This was the most informative video on HDR I've ever watched!
I learned SO MUCH from this video. And still, half of it is over my head. What an awesome video!
Very impressive. This demystifies some misconceptions I've had that I didn't even know needed demystification!
I never learned so much, so quickly, as I do watching your videos ! Thanks so much !
My very first yt comment to say thank you Mr Hess. Inspiring educational technic. Very thorough and entertaining as well. 👏👏👏
Umm John, I am re-watching this video because your explanation of raising the ISO was over my head. If I am not mistaken, or getting crazy, you updated it and added new elements, especially the cloud example. Now I get it. Thanks so much!. This is a great video!. Cheers
Unfortunately can't add or edit TH-cam Videos... else I'd fix a bunch of stuff lol
@@FilmmakerIQ Well, maybe I got confused with one of your other videos, anyway I got it this time. hehehe. Cheers
@@FilmmakerIQ "Unfortunately can't add or edit TH-cam Videos... else I'd fix a bunch of stuff lol"
🙂 Change "TH-cam Videos" to "past actions" and you have my number one frustration with life.
th-cam.com/users/sgaming/emoji/7ff574f2/emoji_u1f642.png
Thanks for the informative (and counter-intuitive!) video, and even bigger thanks for discussing the ISO issue in the comments so patiently. The latter really helped me start to get my head around it.
The whole section of gamma and log is explained so well, thank you!
Best explanation of dynamic range I've heard so far. Great work.
MIND BLOWING...!!!
Lower your exposure, higher ISO to protect highlights, Open up your exposure lower ISO to protect your shadows,.....Am Jumping up and down joyfully for just learning this GEM....Thanks
Hey John ,I love your videos , you teach me somethings that no one do before , and you do it for free . i Wish you the best ❤❤❤
I honestly thought I had a fairly decent grasp on Dynamic Range before, but this video has really shown me so much more
Excellent work! I know nothing about any of these subjects, you are educating me, Thanks
This is one awesome video. I cannot thank you enough; this information was extremely well presented and the presenter was great, too. Thank you so much!!!!
Best video ever on Dynamic Range and log!
Your videos are SO GOOD! SO SO GOOD! =) Total pleasure to watch ads on your video. Totally worth and hope you can continue to make videos like this. Sharing knowledge is what will drive us forward.
ThankYou soo much for this! Watched your previous video and this one and loved both! Your style, teaching method, editing, scientific research is perfect! Look forward to more and like everyone else said the higher iso to protect the highlights and lower for the shadows is blowing my mind
Wow, you are a great teacher. I went from knowing "a lot" about photography, but being baffled (it turns out) by the subjects you covered, to having a basic understanding about the subjects you covered. Thank you.
Great Video and Finally someone who underlines what I said ever since I red an Article on Red's website concerning ISO.
If you want to preserve highlight use a higher ISO and if you want to preserve shadows use a lower ISO. I felt Like a madman trying to explain ppl, how the sensors work, but now I will just link to your video instead! Thank you very much!
And I am looking forward to the HDR Video! Cause that's another topic that noone seems to explain properly. I hop you can explain it!
Wow! This video is trully amazing, I will have to watch more times to get everything! Congratulations for the monumental work!
Great video! I love the depth you go into - I always learn something new, even for subjects I think I understand perfectly well.
This video is worth money. I wasn't even here for photography, but trying to master the gamma & linear concepts.
People used to pay for this kind of knowledge. Thank you for this.
This is a wonderful lesson, well worth the wait since our last one. However, there is just one single thing I'd like to point out regarding BT.709.
The BT.709 specifies a standard that includes a specific color gamut (identical to sRGB), white point (D65) and color space (YCbCr), which then also includes the levels of course (SMPTE Legal Levels). However, it have never really had a specified standard for EOTF though 2.2 have certainly been widely used in this regard. The closest we would get to an EOTF standart for BT.709 would be BT.1886, which was passed in 2011.
I would also like to point out that certain CRTs have very deviant EOTF in comparison to 2.35-2.55, and they also tend to have very variable EOTF through the greyscale. All of this is to be expected of course as the CRT itself is naturally imprecise in this regard. For instance, the Sony BVM-D24E1WU can go almost as low as 3.0 in the shadow detail as its curve gradually goes towards 2.5 at the brightest highlights (though most of the middle grey is certainly at around 2.7-2.8).
Funny enough, the most accurate CRT for I've seen in this regard (with least deviant EOTF curve) is the Sony HDM-3830, which was part of Sonys first HD CRT. It follows Power Law 2.4 almost perfectly, something I've never seen on any other professional grade CRT. Some Ikegamis I've seen have been very close however.
That's a great clarification. Thank you!!!
Thanks so much for doing this! Again one of the best learning experience I have ever had ! Have you ever considered making an episode about color space workflow using ACES?
Yeah color space might be another one to dive into in the future.
You're the best cinema teacher ever!
Yaaas! The professor has returned!!
Finally a video which embraces the most basic topics.
Mate, I think I need to watch it 3 times at least to get a grasp of it. Thanks for doing the grunt work for us.
Respect. Thanks for making this freely available!
What a wonderful surprise to the evening!
Really loved this video! Very informative. Helped me gain a deeper understanding of DR. Hope to keep seeing more videos like this.
The key master and gatekeeper to understanding Dynamic Range :)
No body can explain photography concepts as well as John does! Thank you.
ETTR = Mind blown! The realy good stuff starts at about 24:30
Recién empiezo a estudiar algo de fotografía y no entendía porque muchas paginas me decían que subiera el ISO en situaciones de poca luz cuando por experiencia en post ya me había dado cuenta que era obtenía mas información haciendo lo contrario. Gran explicación!
Information overload! In a good way. I gotta watch this a few more times :)
This was a fantastic episode. One of my favorite you’ve done yet!
Awesome work, I think I'll have physics students watch this video, just taught sound intensity and level (dB) and this would be a good follow on topic!
The advice @26:15 (shoot at higher ISO to preserve highlights) is going to sound counterintuitive to vloggers who are used to shooting ETTR with wide open apertures (to minimize depth of field). Most shooters are familiar with sensor dynamic range charts that show maximum DR occurs at the lowest native ISO settings. If you follow those charts, you'll set ISO at minimum, shutter at 180-degrees, open the aperture as wide as you can, and use an ND filter to prevent the highlights from clipping. That will give you ETTR without reference to middle gray level, which vloggers typically ignore.
The workflow recommended in this video is based on the assumption that you are following the cinematography convention of exposing middle gray at around 38%. The rationale for that practice is not necessarily to maximize DR for each shot, but to establish a common middle gray reference point to make intercuts between footage consistent. With that workflow, raising the ISO will increase the brightness of middle gray, and you will have to lower your exposure (by narrowing the aperture or using more ND) in order to push middle gray down to 38%. It is that adjustment that gives you more highlight range above middle gray. What actually produces more highlight headroom is not simply raising the ISO, but lowering the amount of light that is passing through the lens to the sensor.
This is incorrect - it's fundamentally flawed way of looking at the light. The problem is ND filters and exposures don't attack just highlight range - the apply throughout the range evenly. If you reduce the light to accomodate the highlights, you reduce the light in the mid tones and shadows at the same time.
Yes, raising the ISO will increase the brightness of middle gray. But reducing the exposure to counteract a raise in ISO doesn't actually protect your highlights - it does nothing. Visualize it this way: You have a middle gray and a highlight +5 stops above. You raise your ISO, so now what was middle gray is +1 stop above... now your highlight is +6 stops... so you reduce your exposure... you're back to middle gray with a highlight at +5 stops.
Thus the only thing you did was raise the noise floor by 1 stop if you did that...
This is with the traditional understanding of ISO, but many modern cameras (especially "cinema cameras") utilize a push pull system for ISO where the dynamic range is consistent throughout a chunk of ISO range. This article using my video on the BMPCC 4K is an example of this: www.premiumbeat.com/blog/dual-native-iso-explained/
You'll find similar charts on Red/Arri/Canon cameras as well. Those cameras will allocate more of the dynamic range shades above middle gray as you step up the ISO.
I also explain it all at 24:12 in this video which you should revisit.
Finally middle gray isn't a cinematography convention - it's a photography convention that goes back to the origins. And middle gray is central to the concept of defining ISO as well. Film timing to correct and get consistent exposures has been a thing since forever - it's not why we shoot for middle gray.
I personally would advocate to put the preference to middle gray and then just look out for highlights and shadows that fall out of range and try to wrangle them in. You can try to expose in such a way to capture every bit of dynamic range but I think that misses the forest for the trees.
@@FilmmakerIQ - "But reducing the exposure to counteract a raise in ISO doesn't actually protect your highlights - it does nothing."
Ah, but it does if you reduce exposure by actually lowering the amount of light that passes through the lens (i.e. by narrowing the aperture or increasing the ND filter). That is how you get an extra stop of highlight headroom, by making room for more stops of light intensity above middle gray, not simply by increasing ISO. Notice, I'm not talking about increasing DR here, only about increasing highlight headroom above middle gray, which as you advised, is the point of raising the ISO.
What is actually "non-traditional" about BlackMagic's chart of ISO vs dynamic range is their virtualization of the concept of ISO. The reason the dynamic range of the BMPCC stays constant across a range of ISO's in their chart is because they're not actually recording at incremental analog and/or digital gain levels, they're recording at the native ISO for each range, and using metadata to record the virtual ISO for decoding purposes. In other words, the same range of light intensities (i.e. what is actually detected by the sensor) is being recorded at each ISO step in the range, the only thing that changes is the designated level of "middle gray". But since middle gray is the reference point upon which all other DR measurements are based, everything else changes accordingly.
_but it does if you reduce exposure by actually lowering the amount of light that passes through the lens. That is how you get an extra stop of highlight headroom_ NO IT DOESN'T. You are not giving yourself headroom by changing the exposure, that's thinking by putting the cart before the horse. The light coming in from the lens will have the same dynamic range regardless of ND filters and exposure settings. Proportions between tone ranges are not changing.
It's the ISO that determines the headroom. Adjusting the sensor setting gives you allocation of the highlights and shadows, not the exposure. You match the exposure for the ISO but the exposure is NOT responsible for giving you headroom (unless you purposely underexpose).
In other words - if you just increase the ISO (on these cameras that push/pull for ISO) without changing the exposure - you will increase the headroom above middle gray... but everything will be just overexposed. The headroom above middle gray exists whether you utilize it or not.
I think that is where you confusion is - there is no "middle gray" in terms of light from the scene - it's only in relation to how we set up the sensor.
Vloggers would do best to really understand what I'm saying - what you're saying is just adding to confusion.
@@FilmmakerIQ "The light coming in from the lens will have the same dynamic range regardless of ND filters and exposure settings. Proportions between tone ranges are not changing."
It's not the dynamic range of the light itself, it's the maximum luminance the sensor can detect before it saturates (i.e. the limits of its sensitivity to light), divided by its noise floor, that determines the native dynamic range of the camera. As you point out, a Dual Native ISO camera has two separate sensitivity ranges, but within each range its native sensor sensitivity cannot be changed by ISO settings. Once you reach the maximum luminance that sensor range can handle, the ONLY way you can increase the maximum light level it can discriminate is to lower the amount of light passed through the lens (i.e. by narrowing the lens aperture or increasing the ND filter).
While it's correct to say "it's the ISO that determines the headroom", the BMPCC Dynamic Range vs ISO chart illustrates what the camera's ISO setting is actually determining. Within each sensitivity range, the camera's dynamic range remains unchanged. The ONLY thing that changes when you change the ISO is the designated level assigned to middle gray within each range (until you get to extremely high ISO's). My hat's off to BlackMagic for making it unmistakably clear what the camera's ISO setting actually does.
I think I see where your argument is going but where this is getting flummoxed is you're trying to avoid middle gray.
_Once you reach the maximum luminance that sensor range can handle, the ONLY way you can increase the maximum light level it can discriminate is to lower the amount of light passed through the lens_
Agreed. But that says nothing about headroom. Headroom is defined by the range above middle gray. What is middle gray is ultimately is the question.
Every ISO has a maximum light input... On the BMPCC example ISO 100 an ISO 1000 both have the same maximum light level it can discriminate - ISO 1000 has a lot more headroom between middle gray and full saturation than ISO 100...
But if you were to completely ignore middle gray and worry only about staying just below the clipping point (ETTR)- the idea that reducing the light that lands on the sensor so that's it's further from the clipping point works out as way of "protecting your highlights"
Taken to the extreme, it doesn't matter what ISO you shoot on the BMPCC, so long as you anchor your exposure to just under the clipping point of which ever "native ISO" you chose. The resulting images will be underexposed light scenes and over exposed dark scenes - which you can in theory fix in post.
But I think that approach is fraught with dangers and creates more headaches then it solves actually trying to expose for middle gray.
But honestly, I don't know where this conversation is heading or what you were trying to correct me on. ISO determines dynamic range, you expose for the ISO - I haven't seen anything to counter that line of thinking :)
Amazing video with deep details, but understandable form (if you watch couple times :) ) Thank you for your awesome work here!!
@7:31 I'm really struggling to understand how you get the orange curve. I get lost when you say "Measure the distance between each stop of light". What am I missing? Can you help explain this for me?
Wonderful video and thank you for putting it together.
The Orange curve there is the slope of the power curve. It represents the psychological perception of the next stop of light.
I have two questions which i am slighlty confused about.
1. In stevesn power law there is k which represents the constant of proportionality, but hwat does that actually represent and what number did u use and why when calculating the perceived light sensation.
2. at 7;40- i get how you achieved the first graph by plotting the perceived light sensation on the x ais and teh light intesnity of different sorces on the y axis but what did u mean by teh change in power curve at 7:55-
1. K is just a scaling constant and it changes depending on the units used. For this discussion, all that matters is it's a positive number.
2. Plotting the "change" in the second graph is asking, how much do we "lose in experience" as we step down. From 100% to 50% (1 stop loss), we only lose an experience of "0.2" from 1.0 to 0.8 on the y axis) From 50% to 25% we lose about "0.16" on the y graph - then "1.2" for the next stop.
The point is even though we are reducing the light exponentially, the experience stays relatively linearly constant - it's not flat, but it looks like it evenly steps down in brightness even though in actually light units, it's decreasing geometrically.
Mr Hess, how do I properly express how amazing your videos are and how grateful I am and how grateful I am sure everyone else is? Hmm let me think, you deserve something good. Okay, got it. Here goes: If I were a woman, and if I more specifically were a Jessica Rabbit style femme fatale woman, I'd come over and thank you in the best way, totally rock your world. :D GREAT work!
I don't know what to say except thank you haha
Took me back to 1984 when I constructed my first video frame grabber. I experienced ALL of the issues you discussed here. It was a good ten years later, as a professional broadcast equipment designer, that I caught up with the theory behind it. However, early on, using only my squinty eyeballs and pure experimentation, I created very effective gamma correction values to apply to the look up tables built into the color D/A converters! Even though, I had only the vaguest notion of all the psycho visual factors involved. I loved every second of it! (Don't even get me started on all the nuance of aspect ratios! Pixels vs clock rates vs screen dimensions, etc! Why aren't my calculated circles actually round? ARRRGGGHH!) Thanks!!!!
Thanks for the great explanation which inspires to be dwelve deeper in to this topic..🙏
Canon has this "Save Highlights" option, which basically shoots at one ISO higher than normal and then reduces the exposure in camera one stop. Makes totally sense now.
Another great video. Tons of information, logically presented. Thank you. Thank you. Thank you.
Amazing video. One thing to notice, Sony seems to apply gamma to RAW files in Picture Profiles. There is a video about that on TH-cam.