Thanks a lot. That is a great technical video. Only request for next time would be to add an "total energy chart" where total energy is the sum of the kinetic energy of the ship and its altitude potential energy. This would be very interesting during reentry to see how fast the ship loses its energy.
Nice video. I’m pretty sure however that they did not fuel to 100%, at -T6:48 they show the fuel levels in the tanks fairly close to finishing propellant loading. The fullest tank is the ship methane tank at 91%, while the booster LOX is only 83%.
It's true. My data only starts measuring once the time hits T-, then it updates the fuel. It starts off by default full and then starts reading the fuel bar so the first update to the fuel is the correct update. You are right that it doesn't start full of course!
As others have noted, accurate G values would be good. When starship is falling at more than 250m/s but maintaining the same speed, the g force it's feeling is definitely not 0 (simple physics - Kepler's second law). I hope for the next flight this is included, great video!
Yes, I will try to improve the G-force calculations. Because it runs in real-time, there's only so many calculations I can before it slows down and it takes more than 1 second though so I have to be careful with that.
Nice, the acceleration chart could include gravity. Let's say that when booster is standing on pad, the g should be 1g, and when in freefall, it should be 0g
Indeed, but that would require extra calculation work as well since now I have to calculate it based on orbital velocity as well. I am working on it right now though
Pretty cool visualization. It is very striking to actually see how little fuel is left in the booster at stage separation, and of course even less for the landing burn itself.
Thanks my man. I wanted to add them but getting data on placement/size/volume etc. was difficult. Couldn't make heads or tails of people's estimates so left them off.
booster uses seperate tank for landing, which is present inside the bottom of the lox tank. and starship's header tank depleting animation is not present.
Thanks for the info. The header tanks aren't depleting in the starship because if you notice on their infographics the engines are never shown as re-lighting for the landing burn. My code looks for those graphics to light up first and then it starts tracking and altering the graphics. Unfortunately, since their own graphics don't show it I can't really fix that in any reliable way right now. I hope they fix it for IFT-6!
Very nice, especially for real-time! I did a Python OCR of this as well and wrote a simple filter to get specific force from the engines during the initial Booster burn (the throttle-down for Max-Q is really apparent). I’ve written most of a six-state Kalman nav filter to reprocess it so that I could extract some engine parameters, but that won’t be ready for flight 6. I may steal your fuel tank levels, I hadn’t decided if I was going to try to do that yet. But it would probably be helpful for looking at engine parameters.
Nice man! Sounds great I look forward to seeing it. All this runs in real time and from the X live stream URL, so the plan for this is to livestream it at the same time as the launch. However, I'm working right now on code for in-depth post-launch analysis where every frame of the video is analyzed. Anyway, that'll come later. Thanks though!
Yes I did. I tried my best to compensate for it but it was bad. Some of the acceleration spiking is from missed velocity OCR but some is also from frame drops and other timing issues related to that.
What an awesome project idea! how dependent is the code based on the positioning of onscreen data visuals provided by spaceX? Bravo on the implementation. Also, looks like your sampling rate is ~1Hz, I imagine that the processing time for generating the plot animations would be much much higher if you boosted it, but I imagine 10-30hz would be much more visually appealing.
Indeed, but python isn't super efficient. I would like to make it faster. I could probably push 2 Hz. I plan on doing a deep post-launch analysis at 30 hz, the framerate of the video.
Great job, as an avid KSP RSS player I think you could include approximate TWR from the fuel flow rate tough acceleration is mostly aligned with it (show TWR for 100% ASL and 100% VAC side by side).
@@SpaceLaunchLabs On wikipedia it's 350 (250+100) tons but I think it was 250 + 120 some time ago, it's constantly changing due to changes on Starship. Both values should be really similar until mach one so you will be able to correct it after this estimation.
Interesting... I went through a similar process and pulled data from OCR but have quite different G readings... I understand from comments below that you are running this through filters to clean up the data a bit, but I'd love to know what math you used to generate the G figure. I'll fess up that my first attempt I used math for M/S instead of KM/hr and so was waaay out. Cleaned it up though and so I was surprised to see such a decent delta between our figures.
Hey just so you know I'll explain it when I get back to my computer. I haven't forgotten this comment! If a day passes and I haven't replied with it remind me again!
@@JaivianDean Ah, can you give me a side view schematic of it then? I can incorporate it in the future if I know the blueprints and how much fuel is in the landing tanks.
The acceleration graph is very misleading. There is no direct link between the derivative of the velocity and the g-forces felt by the vehicles (think of a centrifuge, the velocity is constant, but the g-forces are not zero). For example during the boostback phase, the acceleration goes from negative, to zero, to positive, all while the engines are still running, it makes no sense.
Yep, it's based on values on screen so all I have to go on is the velocity shown and I just take the derivative. Obviously I simplified it for the video but it's true that isn't necessarily the true acceleration you would feel in the ship. Without the full velocity vector it's the best I can do.
@@SpaceLaunchLabs You could estimate the velocity vector. The vertical component is the derivative of the altitude (which is known). Then you calculate the horizontal component which would give the correct magnitude of the velocity vector, the latter is of course also known. Of course, for better results you would need to fit a physically possible trajectory to the raw data in order to smooth the noise out.
Funny you say that because my current work is on frame-by-frame analysis of the launch video to do just that! I just need to implement a ship angle detector based on the angle of it in the video...
![Как создаются Микрочипы? Этапы производства процессоров [Branch Education на русском]](http://i.ytimg.com/vi/zyr-I9PdIac/mqdefault.jpg)
![[ ออกกอง ] วัยหนุ่ม 2544 โลกหลังลูกกรง..ทัวร์กองถ่ายหนังใน "คุกของจริง" | JUSTดูIT.](http://i.ytimg.com/vi/vB--aOkZCpw/mqdefault.jpg)
Very well done! Thanks for the hard work! I told a few people to give you a subscription! This needs more reach! ❤️🔥
@@Whataboutit thanks! Love your channel too. Good motivation to keep going!
@@SpaceLaunchLabs Keep that motivation! If you’re open for a collaboration, let me know! You’re doing a great job! @felixschlang on X! ❤
Thanks a lot. That is a great technical video. Only request for next time would be to add an "total energy chart" where total energy is the sum of the kinetic energy of the ship and its altitude potential energy. This would be very interesting during reentry to see how fast the ship loses its energy.
Good idea! I'll consider it!
Fantastic. Note : ship fuel flow wrongly massively drops at th-cam.com/video/LDm1sMPVujc/w-d-xo.htmlsi=wqLxXWvOJnFm1I9Z
At 8:02 video tone
Enjoyed the visualizations. It was cool to see the terminal velocity achieved by Starship when the velocity chart leveled off.
Indeed!
As well as watching it slowly creep down as the air got more dense on descent.
Nice video. I’m pretty sure however that they did not fuel to 100%, at -T6:48 they show the fuel levels in the tanks fairly close to finishing propellant loading. The fullest tank is the ship methane tank at 91%, while the booster LOX is only 83%.
It's true. My data only starts measuring once the time hits T-, then it updates the fuel. It starts off by default full and then starts reading the fuel bar so the first update to the fuel is the correct update. You are right that it doesn't start full of course!
this is awesome! can't wait for IFT-6's data too
Great video. Puts into perspective how much energy is needed to put anything in orbit.
As others have noted, accurate G values would be good. When starship is falling at more than 250m/s but maintaining the same speed, the g force it's feeling is definitely not 0 (simple physics - Kepler's second law). I hope for the next flight this is included, great video!
Yes, I will try to improve the G-force calculations. Because it runs in real-time, there's only so many calculations I can before it slows down and it takes more than 1 second though so I have to be careful with that.
I cant wait for the ift 6! If you manage to do it in a short time, i would be really suprised.
When will the live stream begin? Excited for flight 6!
When the SpaceX livestream starts! The beginning will just be me re-streaming it and then when it hits T-00:05:00, the data will appear.
WOW. I find this mighty impressive and a superb 'tech demo' on OCR and thinking out of the box. Kudos ! Nicely done.
Glad you like it!
I would love to see a similar chart for Falcon 9 (Heavy)
Support my work and it can be done! The basic building blocks are done and could be ported over to Heavy streams.
Can't wait to watch this during ift-6. this is all very interesting to look at
it is crazy how much those engine can throttle almost 10 times from lowest to highest output
Nice, the acceleration chart could include gravity. Let's say that when booster is standing on pad, the g should be 1g, and when in freefall, it should be 0g
Indeed, but that would require extra calculation work as well since now I have to calculate it based on orbital velocity as well. I am working on it right now though
Pretty cool visualization. It is very striking to actually see how little fuel is left in the booster at stage separation, and of course even less for the landing burn itself.
Yes the bar gives a hint of that but seeing that sliver at the bottom is quite jarring!
Very interesting! Thanks for this!
My pleasure!
only thing I'd add is booster Header Tanks.
This is crazy man. Nice work!
Thanks my man. I wanted to add them but getting data on placement/size/volume etc. was difficult. Couldn't make heads or tails of people's estimates so left them off.
that's a nice work you did!
Appreciate it!
amazing work, this is what i wanted to see.
Thanks!
booster uses seperate tank for landing, which is present inside the bottom of the lox tank. and starship's header tank depleting animation is not present.
Thanks for the info. The header tanks aren't depleting in the starship because if you notice on their infographics the engines are never shown as re-lighting for the landing burn. My code looks for those graphics to light up first and then it starts tracking and altering the graphics. Unfortunately, since their own graphics don't show it I can't really fix that in any reliable way right now.
I hope they fix it for IFT-6!
Oh this is amazing!!! great work. Thanks for breaking down the data like this.
Glad you liked it!
Awsome! Id probably average the acceleration data out a bit bc it was quite choppy. I really wish companies would give us data like this. great work!
Yep, I actually run it through a kalman filter but without more precise time codes it's the best I can do.
Sooo, if I understood well, you did all these datas with only the data on the livestream screen of SpaceX, is that correct ? Amazing work !!!
Yep, that's correct!
Very nice, especially for real-time! I did a Python OCR of this as well and wrote a simple filter to get specific force from the engines during the initial Booster burn (the throttle-down for Max-Q is really apparent). I’ve written most of a six-state Kalman nav filter to reprocess it so that I could extract some engine parameters, but that won’t be ready for flight 6. I may steal your fuel tank levels, I hadn’t decided if I was going to try to do that yet. But it would probably be helpful for looking at engine parameters.
Nice man! Sounds great I look forward to seeing it.
All this runs in real time and from the X live stream URL, so the plan for this is to livestream it at the same time as the launch. However, I'm working right now on code for in-depth post-launch analysis where every frame of the video is analyzed. Anyway, that'll come later. Thanks though!
@ Did you notice the frame instability in the Flight 5 stream replay as compared to Falcon 9 replays? It was really bad.
Yes I did. I tried my best to compensate for it but it was bad. Some of the acceleration spiking is from missed velocity OCR but some is also from frame drops and other timing issues related to that.
@ it’s not letting me tag you in my response with the plot I made. Can you see the comment and the link?
No I can't unfortunately. If you have X DM me there
@LaunchLabsLive
What an awesome project idea! how dependent is the code based on the positioning of onscreen data visuals provided by spaceX? Bravo on the implementation. Also, looks like your sampling rate is ~1Hz, I imagine that the processing time for generating the plot animations would be much much higher if you boosted it, but I imagine 10-30hz would be much more visually appealing.
Indeed, but python isn't super efficient. I would like to make it faster. I could probably push 2 Hz. I plan on doing a deep post-launch analysis at 30 hz, the framerate of the video.
I wish I could code. Great job! I'll be back on Tuesday. Subbed. :)
Thanks!
Why do we get negative g readings on descent for Starship? Reference frame has changed?
@@jwandel I could've taken the absolute acceleration but I left it as negative to make it clear it's decelerating. No reference change
@ yeah I guess I meant the other way around. Precisely the reference frame did NOT change. Good point, cool, and thanks for the vid.
Very interesting
Thx for the video
I find it interesting that you hadn't uploaded in 15 years and you post again on this channel just yesterday with something completely different.
Hehe, I was but a teenager back then!
Wow what a video. How did you manage to do this? Very well done.
Thanks! It's from the on-screen information, using only video OCR.
Great job, as an avid KSP RSS player I think you could include approximate TWR from the fuel flow rate tough acceleration is mostly aligned with it (show TWR for 100% ASL and 100% VAC side by side).
I can do that. What's the dry mass of both booster and starship?
@@SpaceLaunchLabs On wikipedia it's 350 (250+100) tons but I think it was 250 + 120 some time ago, it's constantly changing due to changes on Starship. Both values should be really similar until mach one so you will be able to correct it after this estimation.
Vehicles do not start with a full propellant load since IFT-3
I know, it takes the OCR a bit to capture the current load. The first time it updates is the correct propellant load.
Absolutamente insano.
Awesome video!!
Thanks man!
5:20 ~ 5:23 looks like the starship attitude data have a glitch?
also free methane 9:56 xd
Hehe, yeah OCR can be finicky. I could mess with it so it captures it perfectly but it wouldn't run in real time any more!
How did you get this data or do you work for SpaceX?
It's OCR from the video, based entirely on just the on-screen info.
Interesting... I went through a similar process and pulled data from OCR but have quite different G readings... I understand from comments below that you are running this through filters to clean up the data a bit, but I'd love to know what math you used to generate the G figure. I'll fess up that my first attempt I used math for M/S instead of KM/hr and so was waaay out. Cleaned it up though and so I was surprised to see such a decent delta between our figures.
Happy to share some data on google sheets if you like btw
Hey just so you know I'll explain it when I get back to my computer. I haven't forgotten this comment! If a day passes and I haven't replied with it remind me again!
Can't DM via youtibe, do you have any other socials?
Will be following tomorrow!
Also, it seems out graphs some very similar curves and so it's just a slight misalignment of the metrics
I have an X account: LaunchLabsLive
DM me there!
.5g during liftoff???
hello it look really cool can you make it open source please! it look really sick
I will eventually once the code is more mature and user friendly.
In the booster fuel map, did you account for the header tanks?
From my understanding the booster doesn't have header tanks.
@@SpaceLaunchLabs It has landing tanks though
@@JaivianDean Ah, can you give me a side view schematic of it then? I can incorporate it in the future if I know the blueprints and how much fuel is in the landing tanks.
That is VERY interesting
Thanks!
Booster Header?
Couldn't find enough info on placement/size so didn't include it unfortunately.
And every SLS engineer updated their resumes…
The acceleration graph is very misleading. There is no direct link between the derivative of the velocity and the g-forces felt by the vehicles (think of a centrifuge, the velocity is constant, but the g-forces are not zero).
For example during the boostback phase, the acceleration goes from negative, to zero, to positive, all while the engines are still running, it makes no sense.
Oh yeah that is a little weird
I dont they have the actual g-force data though, i think they are coding it based off the values shown on screen.
Yep, it's based on values on screen so all I have to go on is the velocity shown and I just take the derivative. Obviously I simplified it for the video but it's true that isn't necessarily the true acceleration you would feel in the ship. Without the full velocity vector it's the best I can do.
@@SpaceLaunchLabs You could estimate the velocity vector. The vertical component is the derivative of the altitude (which is known). Then you calculate the horizontal component which would give the correct magnitude of the velocity vector, the latter is of course also known. Of course, for better results you would need to fit a physically possible trajectory to the raw data in order to smooth the noise out.
Funny you say that because my current work is on frame-by-frame analysis of the launch video to do just that! I just need to implement a ship angle detector based on the angle of it in the video...
Is this the real data from the rocket or just phony stuff?
It's real from on-screen data!
20% of booster fuel in 20 secs 💀💀💀💀💀💀
От 0 км/ч до скорости звука энергетическая яма которую нужно пролететь максимально быстро.