Author Topic: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread  (Read 964048 times)

Offline Swatch

  • Full Member
  • **
  • Posts: 273
  • Official Aerospace Engineer as of June 13th, 2009
  • Cincinnati
    • ProjectApollo/NASSP: Virtual Systems and Flight Simulation of the Apollo Program
  • Liked: 28
  • Likes Given: 16
Chris, may I recommend you add links to the first post of this thread detailing how to start helping with this effort and some of the better posts in the thread describing what's going on?

I think it will help orient people without going through 24 pages of stuff.

I'd rather people read through what is only 24 pages. It takes 5-10 minutes to read and new people to the thread will gain an appreciation of the work and the processes involved. :)

Maybe just a link to the wiki page then?
Ex-Rocket Scientist in Training, now Rocket Scientist!
M-F trying to make the world of the future a smaller place through expanding horizons...

Offline MarsInMyLifetime

  • Full Member
  • **
  • Posts: 278
  • Austin, TX
    • Light of Day
  • Liked: 132
  • Likes Given: 137
Chris, may I recommend you add links to the first post of this thread detailing how to start helping with this effort and some of the better posts in the thread describing what's going on?

I think it will help orient people without going through 24 pages of stuff.

I'd rather people read through what is only 24 pages. It takes 5-10 minutes to read and new people to the thread will gain an appreciation of the work and the processes involved. :)

Maybe just a link to the wiki page then?
I think he's saying that if you read the pages, you'll come across the link anyway, but by then you'll have much better context and appreciation for the incredible effort going on. As a Web publisher myself, I have to keep in mind (and be reminded) that I want to try to keep visitors ON my site for as long as I can. Besides, the best exhilaration is when you keep coming across ever new news here, right?
Don

Offline ugordan

  • Senior Member
  • *****
  • Posts: 7617
    • My mainly Cassini image gallery
  • Liked: 1840
  • Likes Given: 406
I don't know how much "better" such video would be. On another site I saw a claim that the telemetry plane on CRS-3 was over 20 miles away. The CASSIOPE launch aerial images released also suggest a pretty large distance from the vehicle. I guess safety trumps closeup observations.

There was really bad weather, so ships and planes that were slotted to be much closer to the "action" weren't able to get there for safety reasons. 

SpaceX couldn't have possibly asked for *better* weather for the CASSIOPE launch and, yet, the photography of the 1st stage was still very distant and blurry.

Offline Swatch

  • Full Member
  • **
  • Posts: 273
  • Official Aerospace Engineer as of June 13th, 2009
  • Cincinnati
    • ProjectApollo/NASSP: Virtual Systems and Flight Simulation of the Apollo Program
  • Liked: 28
  • Likes Given: 16
...
Maybe just a link to the wiki page then?
I think he's saying that if you read the pages, you'll come across the link anyway, but by then you'll have much better context and appreciation for the incredible effort going on. As a Web publisher myself, I have to keep in mind (and be reminded) that I want to try to keep visitors ON my site for as long as I can. Besides, the best exhilaration is when you keep coming across ever new news here, right?

Duly noted... but I was trying to be lazy and have had trouble finding the link again for myself with it buried in the thread. :P 

Can't always spend a lot of time digging through the thread every time I want to see the updated pics when  there are more rockets to build!  ;)  Things are hectic around here, but it's amazing following the stuff you guys are doing with this video.  :)
« Last Edit: 05/08/2014 07:57 PM by Swatch »
Ex-Rocket Scientist in Training, now Rocket Scientist!
M-F trying to make the world of the future a smaller place through expanding horizons...

Offline arnezami

  • Full Member
  • **
  • Posts: 284
  • Liked: 262
  • Likes Given: 346
Hi guys,

This is just for inspiration:



"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Offline CraigLieb

  • Full Member
  • ****
  • Posts: 872
  • Dallas Fort Worth
  • Liked: 805
  • Likes Given: 1003
Hi guys,

This is just for inspiration:

"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Terrific video.

And also useful to note the side view of the rocket descending into the smoke,
 much like the stage returning as it hits the surface of the water.

I suppose this is obvious to the rocket scientists that we would see just parts of the struts.
« Last Edit: 05/08/2014 08:49 PM by CraigLieb »
Colonize Mars!

Offline grythumn

  • Full Member
  • **
  • Posts: 221
  • Liked: 129
  • Likes Given: 154
More cleanup of iframe 9. These are from try1, but not much changed at least in the top of the frame...

Now that iframe 8 is almost done, pulled the clock position from it and set the top of this clock to that... gives us a reference point for the bottom of the frame.

X:16410:01,X:19164:10,X:19903:1,X:20037:80,X:27244:1,X:32743:1,X:32551:80,20:04:-1,
24:04:16023,16:5:-1,17:05:18692,24:08:-1,25:08:32410,1:15:-1,3:16:79127,41:19:-1,8:20:96220,
15:04:16023,4:28:112147

Bonus animation from when I was testing individual fixes against try1.ts from the old frame.

-Bob

Offline John_L

Just curious...
The original releases (raw and 'repaired') had frames as the stage was descending through the cloud layer which fit before this sequence, and the stage falling over at the end which come after -- both best seen in raw IMO.  Are those frames somewhere in this amazing recovery effort?

The answer is yes and no...

MPEG encoding has different types of frames... I-Frames, which contain All of the information needed to reconstruct that frame, the other types B and P frames only contain partial information, just information about what's changed since the last i-frame.  In our example here, we have a 30 second video with about 307 frames (I think), and of those 307 frames, only about a dozen or so are I-Frames.  Those are the frames we are working on, because the rest of the frames will be generated from data within the I-Frames. A given block of pixels on the screen can rely on blocks before it, after it, or from the I-Frame that was several frames back.

When you see us talking about frame 1, frame 2, frame 3, etc, we're not talking about the first 3 frames of the video, we're talking about just the iframes.  There are (by my guess) about 20-25 intermediate frames between each iframe.  We'll be getting to those once the iframes are fixed, at that point, we'll re-encode the video replacing the bad iframes with the good ones, and once the video is re-encoded, the intermediate frame data will be regenerated from the repaired iframes, and hopefully we'll have a much cleaner video.

Moral of the story... MPEG cheats... ALOT !

I tried not to get too technical with the explanation, hope that clears it up for you though.

John
« Last Edit: 05/08/2014 09:04 PM by John_L »

Online mlindner

  • Software Engineer
  • Senior Member
  • *****
  • Posts: 2090
  • Space Capitalist
  • Silicon Valley, CA -- previously in Ann Arbor, MI
  • Liked: 748
  • Likes Given: 289
More cleanup of iframe 9. These are from try1, but not much changed at least in the top of the frame...

Now that iframe 8 is almost done, pulled the clock position from it and set the top of this clock to that... gives us a reference point for the bottom of the frame.

X:16410:01,X:19164:10,X:19903:1,X:20037:80,X:27244:1,X:32743:1,X:32551:80,20:04:-1,
24:04:16023,16:5:-1,17:05:18692,24:08:-1,25:08:32410,1:15:-1,3:16:79127,41:19:-1,8:20:96220,
15:04:16023,4:28:112147

Bonus animation from when I was testing individual fixes against try1.ts from the old frame.

-Bob

One key thing to note about the clock. The blocks for it are abnormally large so corruption chance increases. They average around 1000 bits each for me.

Also, every timestamp in the video is of the form: 19:35:XX.XX
Importantly, the lower portion of the colons are entirely in the lower two sub-blocks of the bottom row and the upper portions are entirely in the lower two sub-blocks of the second from bottom row. Further, for the "." character its slightly wider than the lower portion of the ":" meaning you can recognize where to put it easier.
Internal combustion engine in space. It's just a Bad Idea.TM - Robotbeat

Offline Alpha Control

  • Veteran
  • Full Member
  • ****
  • Posts: 1195
  • Washington, DC
  • Liked: 140
  • Likes Given: 74


I think it is likely that it could take weeks or months to get the video to a level of quality that is going to satisfy everyone working on it. Chris, can you ask PAO what timeframe would be most useful to them?

You guys are ahead of the game, the only real game in town actually. Provide what you can, when you can. This is an unbelievable effort on show here.

This is truly an d unbelievable effort. And is amazing to see the amount of details that have been recovered form the initial frames, and the conviction and talent of all people that have been involved....

This is what we have so far guys! And let's keep with the good job:
Great! I was about to ask for something like this. Or make it myself. :)
One suggestion/request: a bit slower sequence? Either more realistic in terms of actual movie speed or even slower to be able to find common blocks?

Sure. How about this:

Moralec, that's terrific. Thanks for adding the frame numbers and slowing down the playback.

Really great summary of where this amazing effort is at this point! It's not my field, but it is wonderful to follow (if only at a high level) the great progress everyone is making. 
Space launches attended:
Antares/Cygnus ORB-D1 Wallops Island, VA Sept 2013 | STS-123 KSC, FL March 2008 | SpaceShipOne Mojave, CA June 2004

Offline moralec

Hi guys,

This is just for inspiration:

"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Terrific video.

And also useful to note the side view of the rocket descending into the smoke,
 much like the stage returning as it hits the surface of the water.

I suppose this is obvious to the rocket scientists that we would see just parts of the struts.

The only problem is that in iframe 13, you can see both being the same size again....


Weird....

http://spacexlanding.wikispaces.com/

Offline Asmegin

  • Member
  • Posts: 35
  • Canada
  • Liked: 21
  • Likes Given: 3
A bit more of the Frame 9 clock;

5:28:110675,6:29:120015,4:29:117503

Offline AncientU

  • Senior Member
  • *****
  • Posts: 6249
  • Liked: 4105
  • Likes Given: 5619
Just curious...
The original releases (raw and 'repaired') had frames as the stage was descending through the cloud layer which fit before this sequence, and the stage falling over at the end which come after -- both best seen in raw IMO.  Are those frames somewhere in this amazing recovery effort?

The answer is yes and no...

MPEG encoding has different types of frames... I-Frames, which contain All of the information needed to reconstruct that frame, the other types B and P frames only contain partial information, just information about what's changed since the last i-frame.  In our example here, we have a 30 second video with about 307 frames (I think), and of those 307 frames, only about a dozen or so are I-Frames.  Those are the frames we are working on, because the rest of the frames will be generated from data within the I-Frames. A given block of pixels on the screen can rely on blocks before it, after it, or from the I-Frame that was several frames back.

When you see us talking about frame 1, frame 2, frame 3, etc, we're not talking about the first 3 frames of the video, we're talking about just the iframes.  There are (by my guess) about 20-25 intermediate frames between each iframe.  We'll be getting to those once the iframes are fixed, at that point, we'll re-encode the video replacing the bad iframes with the good ones, and once the video is re-encoded, the intermediate frame data will be regenerated from the repaired iframes, and hopefully we'll have a much cleaner video.

Moral of the story... MPEG cheats... ALOT !

I tried not to get too technical with the explanation, hope that clears it up for you though.

John
Thanks for the generous explanation and all effort on this reconstruction.
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline John_L

Interesting video about how MPEG works...




Offline AncientU

  • Senior Member
  • *****
  • Posts: 6249
  • Liked: 4105
  • Likes Given: 5619
Hi guys,

This is just for inspiration:

"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Terrific video.

And also useful to note the side view of the rocket descending into the smoke,
 much like the stage returning as it hits the surface of the water.

I suppose this is obvious to the rocket scientists that we would see just parts of the struts.

The only problem is that in iframe 13, you can see both being the same size again....


Weird....

http://spacexlanding.wikispaces.com/
The stage has bobbed back up to the surface (mostly) and just has the thrust structure underwater.
The thin 'legs' are actually just the pistons.
« Last Edit: 05/08/2014 09:24 PM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline moralec

Hi guys,

This is just for inspiration:

"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Terrific video.

And also useful to note the side view of the rocket descending into the smoke,
 much like the stage returning as it hits the surface of the water.

I suppose this is obvious to the rocket scientists that we would see just parts of the struts.

The only problem is that in iframe 13, you can see both being the same size again....


Weird....

http://spacexlanding.wikispaces.com/
The stage has bobbed back up to the surface (mostly) and just has the thrust structure underwater.
The thin 'legs' are actually just the pistons.

Yep, that's possible...

The answer lies in the b and p frames :P

Offline saliva_sweet

  • Full Member
  • ****
  • Posts: 572
  • Liked: 446
  • Likes Given: 1428
Ocean from iframe 6 (try1). Unfortunately I have yet to think of a straightforward way to convert between the two framesets. So I can't combine this with other peoples progress. A chain file to tell which bit ranges in original frames correspond to which ranges in try1 would be of great value.
The data in this frame seems to be in fairly good condition. I think most of the frame can be perfectly recovered.

mmb 0:0:583,39:0:5496,43:0:-1,2:1:6129,21:1:8584,22:1:8796,6:3:
18105,7:3:18276,9:3:29527,30:3:32359,31:3:-1,43:3:33591,2:6:44738,4:6:46341,5:6:48794
« Last Edit: 05/08/2014 09:49 PM by Chris Bergin »

Offline MTom

  • Full Member
  • ***
  • Posts: 378
  • EU / Hungary
  • Liked: 107
  • Likes Given: 435
Hi guys,

This is just for inspiration:

"We've always defined ourselves by the ability to overcome the impossible"

Regards,

arnezami

Terrific video.

And also useful to note the side view of the rocket descending into the smoke,
 much like the stage returning as it hits the surface of the water.

I suppose this is obvious to the rocket scientists that we would see just parts of the struts.

The only problem is that in iframe 13, you can see both being the same size again....


Weird....

http://spacexlanding.wikispaces.com/

IMHO:
On frames 10-11-12 you see nearly anything about the legs because of the vapor, likely not because the legs are still partially under the water.
On frame 13 the pistons are much clearer to see - the vapor disappeared.
About from this point are the legs only partially under water (or a little bit earlier, while the rocket should cause also a big wave - it has about 4.5m diameter - this delays also a little bit that the pistons go under water).


Edit: the explanation from AncientU is also logical - the video will hopefully give an answer.

Edit2: If I see the timeline, frame13 is min. 3 seconds after engine shut down. For my theory this is too long time (and in 8 seconds it went horizontal)...
« Last Edit: 05/08/2014 10:08 PM by MTom »

Offline grythumn

  • Full Member
  • **
  • Posts: 221
  • Liked: 129
  • Likes Given: 154
More cleanup of iframe 9. These are from try1, but not much changed at least in the top of the frame...

Now that iframe 8 is almost done, pulled the clock position from it and set the top of this clock to that... gives us a reference point for the bottom of the frame.

X:16410:01,X:19164:10,X:19903:1,X:20037:80,X:27244:1,X:32743:1,X:32551:80,20:04:-1,
24:04:16023,16:5:-1,17:05:18692,24:08:-1,25:08:32410,1:15:-1,3:16:79127,41:19:-1,8:20:96220,
15:04:16023,4:28:112147

Bonus animation from when I was testing individual fixes against try1.ts from the old frame.

-Bob

One key thing to note about the clock. The blocks for it are abnormally large so corruption chance increases. They average around 1000 bits each for me.

Also, every timestamp in the video is of the form: 19:35:XX.XX
Importantly, the lower portion of the colons are entirely in the lower two sub-blocks of the bottom row and the upper portions are entirely in the lower two sub-blocks of the second from bottom row. Further, for the "." character its slightly wider than the lower portion of the ":" meaning you can recognize where to put it easier.

Add ,3:29:118925 for the damaged bottom portion of the clock. The 9 comes and goes, it's pretty damaged.

-Bob

For those that aren't aware there's a new improved online editor that has all the UI for the various options and the ability to use different sources ("coerced.ts", "try1.ts") you can get to it via the wiki here http://spacexlanding.wikispaces.com/ (Online Editor v2)

If anyone has any other .ts they are using I can upload that too, just let me know (I just need the .mp4-img files and a name for it)

Tags: