How to Optimize HDRi's in Blender!
ฝัง
- เผยแพร่เมื่อ 27 มิ.ย. 2024
- High resolution HDRi's use a lot of memory and can slow down the performance of Blender. In this quick tip I want to show you how you can get really good results by combining two HDRi's with relatively low file sizes.
497 Essential Tips for Blender: gumroad.com/l/lDLmX
Blender Daily Patreon: / blenderdailytips
Blender VFX course: gum.co/vfx-course
Blender Daily Gumroad: gumroad.com/blenderdaily
Free Resources: blenderdaily.xyz/free
Video Instructor: Nik Kottmann
Instagram: / nikkottmann
Blender Daily on Social Media
Instagram: / blender.daily
Twitter: / blenderdaily
Facebook: / blender.daily
TikTok: / blender.daily
Website: blenderdaily.xyz/
💥If you like short Blender tips like the one in this video you should check out the collection I created in collaboration with the Blender Bros where we share 497 of our most useful tips & tricks for Blender artists!💥 gumroad.com/l/lDLmX/497
underrated channel . ! love your videos !
hey thanks Hercules
Underrated channel! Amazing video
thank you for your nice words Derny!
Thanks man really appreciate this you are massively underrated - keep it up champ 💯
So brilliant, yet so simple. Perfect for most scenarios. Thanks!
this is an exceptional analysis - simple, elegant, well explained and thorough
As usual, really clever tips, thank you Nick !
hey thanks Alexandre;)
Great tip!
Great Tutorials Nick. Keep it up.
thank you!
Thanks for this very helpful tutorial and happy blending with blender.
Thats huge, thanks man!! keep up the good work
Such a good tip! Thanks!
This worked wow. Another great technique to improve my scene. Thank you!
wow! thanks for this time saving tips!! you're the best!
I've just found this video. Man, you are a genius. Just like this, simple and straightforward: a genius. The real good is that you can modify the brightness and exposure without burning out the background picture. Very elegant solution. Thank you.
simple and perfect. thank you!
Thank you so much. Very useful tip!
dude really nice tutorial... keep up the great work
hey thanks Jorge!
Thanks for this.
Wow, this is really useful. Thanx for the tutorial.
Absolutely great. Thank you!
Thank you, very useful!
thanks nick!!
That is a great tip. In the past I changed the look of the final image to allow use of the 1k HDRI only by using a shallow depth of field to get rid of the pixelation. Depending on what you are looking for that is another alternative.
sounds like a good idea👌🏻
Also if you change the interpolation to Cubic, the blur gets way more even and less pixelated.
@@aeraglyx that is good to know. Thanks for the tip.
brilliant solution!
Nice! Thank you.
Soooo creative 👍
Ty so much big help
How i never remember this? Thank you ;)
yeah it is really powerful:)
Wow thats Genius man!
Thank you brother
PERFECT!!!
very well explained. Thank you. #blender #3dlighting
you just saved my life
Looks like using 1K EXR for lighting and 8K shadow mapped JPG for the actual background image would be the best approach.
Thanks
I think most people dont use the HDRi for background but need sharp and high-resolution REFLECTIONS, that you somehow missed out.
so he should make another mix shader node to use 8k jpeg for reflections right?
@@mcan-piano4718 Well it depends on what you are up to. For example I render hi-res (like 15K) print visuals of cars, and for that a 8K jpeg background really wont cut it, and the reflections of a 1K HDRi on the carpaint would be simply awful. So I rather use actual RAW file photos for the background, composited in in post, and skip this nonsense. What is more important is optimizing the HDRi itself.
blender docs have a correct trick in bottom on "world settings" theme in search
thanks today i have learned something new:)
nice to hear that you find it valuable:)
you're a god
for a moment there i thought your brother was Joji... was expecting the Pink Guy to jump out any minute
Great solution...I have question, how to adjust/rotate the background and HDRI shadow/source lighting? Thanks!
...back in the days we actually used 3 maps... a very low res blurred hdri for lighting+a spotlight for sharp shaddows (reduces sampling amount...although thanks to MIS sampling its less of a problem today), a mid res hdri for reflections, and a tonemapped background or backplates for the scene view.
www.hdrlabs.com/sibl/index.html
Hello. Thanks for your tutorials!!. Very nice . I have a question. After HDRI_HAVEN changed to "Poly Haven" I can not use the addon Lily Surface and its not possible to find the different kinds of formats . Hope you can help me.
We can combine the 8k tonemapped and 1k with camera ray to mix with
Why would anyone dislike this?
I did a bit of simple testing with just a default cube, and by simply watching the memory numbers on the status bar. I also made a 8K jpg completely white. The file size of the jpg is only 100K.
Here are my result when rendering (card is 3GB 1060ti):
8K HDRI: peak memory at 1.54GB, then down to 800 MB, Vram 1.7GB
8K jpg (6MB) + 1K HDRI: peak memory at 1.17GB, then down to 450 MB, Vram 1.3GB
white 8K jpg (100K) + 1K HDRI: peak at 700MB down to 450MB, Vram 1.3GB
So it's definitely a good idea and an improvement but I believe it is important to realise that the file size on disk is irrelevant because the image needs to be decompressed in memory. For ram, the peak memory saving corresponds roughly to the 32 vs 24 bits of resolution between jpg and hdri. Vram is reduced by half which is great. And note that a 100K vs 6MB jpg image makes almost no difference. They use the same amount of memory since they have the same number of pixels.
Conclusion: great idea that I will use. However it is far from a 12:1 memory usage saving that the file size on disk seems to suggest.
And I subscribed
so cleverrrrrrrrrrrrrrrrrrrrrrrrrrrrr so what do u recommenddd
@@mcan-piano4718 Well, my point was that you should not consider the file size in calculations. 2 jpg files (or 2 HDRI files) of the same dimensions will use exactly the same amount of space in memory, regardless of the file size they use on disk.
However HDRI files definitively take more memory than jpg so his trick is a good idea. You can also use the HDRI-sun-aligner add-on that he mentions as part of this tutorial. This saves even more memory because you don't use an HDRI file at all. plus you can tweak the sun a bit. th-cam.com/video/Aco0UDxL6rA/w-d-xo.html
But my biggest recommendation is to upgrade your card if you can. Few weeks ago I purchased a new PC with an RTX 3060 12GB. The difference with the GTX 1060 3GB, combined with Cycles-x is incredible. I have a scene that went from 95 minutes to render down to less than 5. However Cycles-x seems to be using a bit more memory.
Good tip but as a beginner I really think this could have been condensed to 3 minutes or less for the overview and tip itself of combining the nodes. Thanks for the video, much appreciated but but please can you try and condense things next time, the pacing is very slow and I had to watch this at x2 speed to get past pauses in speech etc.
blender docs have a correct trick in bottom on "world settings" theme in search work with hi res reflections
hey I didn't delete your comment. TH-cam automatically removes comments than include links
@@BlenderDaily sorry. i didnt knew.
@@BlenderDaily reupload video with new data
Can u pls post avid on light path node explained
A tonemapped JPG file is a image to view on your monitor. It's like a preview of the EXR "light-data" file that you can download to use for image-based lighting. The EXR file is not an image!
thank you for your insights Daniel!
Thanks. Nice to know that
Why does my HDRI maps that I made doesn't cast a shadow.Im using Theta SC2 I didn't shoot in raw cause it doesn't support but I shoot bracketing exposure. can you help me?
Your HDRi needs to have a high enough dynamic range in order to cast shadows. I explain it in this video: th-cam.com/video/Aco0UDxL6rA/w-d-xo.html
This guy fucking RIPS!!
do you have a discord server for blender too?
I don't have my own server but I have a channel on Daniel Krafft's server:) discord.gg/danielkrafft
@@BlenderDaily yeah i joined it:) and i also follows daniel kraft!
But the small file size of the .jpg on the disk is achieved by compression. Doesn't Blender need to decompress the image into memory in order to use it? In that case the 8K image which is 32 megapixels is still using about 100MB of memory. Right?
It is compressed but it stays at 8 MB. It has no dynamic range though.
Jpeg allways area allways smaller than png, hdr and others, specially because of the small dinamic range
i dont think blender decompresses any type of already compressed images.. jpeg is an 8bit only , 8k image for an 8bit is still low on memory, even though it can give you 32 megapixels, those megapixels aren't dense enough to cover the reflection of the object, because of the 8bit color scale (from black to white) the black pixels are located on the surface , for black being 0 megapixel (as shown in the video). hdr and exr are 32 bit. therefore the megapixels are more dense covering the object's reflection, no black pixels are located which means more emission so more memory.
@@luciox2919
Color jpg are 24 bits since they must use 8 bits for each RGB channel. I posted another message (above) where I kind of prove that the image must be decrompressed in memory to be used. 2 jpg files with 8K resolutions with one file being 60 times larger than the other use the same amount of memory, both in ram and vram.
Is this the technique Blendeguru use in his add on Pro-lighting Skies?