Author Topic: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread  (Read 954655 times)

Offline mhenderson

  • Member
  • Posts: 69
  • USA
  • Liked: 101
  • Likes Given: 18
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1700 on: 06/21/2014 12:18 PM »
Final request -

Please PM me with a rough guess at the number of hours you have spent working on this little labor of love since it began in April. 

I have heard from a decent number of contributors, but more is always better.

I can handle vague.  Range estimates are fine - 0-25, 25-50, 50-100, etc.  Use your gut ... "five hours a week plus or minus over six weeks".  "Two big weeks of 20 and a few weeks of 2-4s".  Even if you never submitted anything, tell me your story "I played with the online editor for three hours but it is over my head." Anecdotal evidence still adds to the picture.

The purpose is to paint a picture in broad strokes of the effort. Information will remain anonymous, only aggregate stats will be revealed. 


Offline MTom

  • Full Member
  • ***
  • Posts: 378
  • EU / Hungary
  • Liked: 107
  • Likes Given: 435
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1701 on: 06/21/2014 12:25 PM »
Final request -

Please PM me with a rough guess at the number of hours you have spent working on this little labor of love since it began in April. 

I have heard from a decent number of contributors, but more is always better.

I can handle vague.  Range estimates are fine - 0-25, 25-50, 50-100, etc.  Use your gut ... "five hours a week plus or minus over six weeks".  "Two big weeks of 20 and a few weeks of 2-4s".  Even if you never submitted anything, tell me your story "I played with the online editor for three hours but it is over my head." Anecdotal evidence still adds to the picture.

The purpose is to paint a picture in broad strokes of the effort. Information will remain anonymous, only aggregate stats will be revealed.

It could happen that earlier contributors don't read this thread actually. Maybe a PM for them would be a good idea.

Offline Quialiss

  • Member
  • Posts: 75
  • Liked: 82
  • Likes Given: 29
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1702 on: 06/21/2014 02:26 PM »
Note for converting i frames:  Don't be surprised if you can/have to get rid of DC changes in macroblocks that are just below a gap.  DC fixes for contiguous data is a matter of finding flipped bits, but correcting for gaps is more a matter of adjusting to compensate for what's missing/smeared out.  Ignoring that data with -2/3 sometimes means less total corrections needed.

Offline aero

  • Senior Member
  • *****
  • Posts: 2974
  • 92129
  • Liked: 793
  • Likes Given: 281
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1703 on: 06/21/2014 03:30 PM »
Hey people, thanks for all the good work.

I just noticed that the latest You tube video on Twitter is two days old, Thursday 2014/6/19 13:02:58 GMT.

Is that intentional?
Retired, working interesting problems

Offline Lindblue

  • Member
  • Posts: 10
  • Germany
  • Liked: 2
  • Likes Given: 1
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1704 on: 06/21/2014 05:37 PM »
When I saw "your" video during yesterdays SpaceX webcast I was really happy!
Been following this thread for months and have to say that you all do fantastic work!
Really impressive.

Offline wronkiew

  • Full Member
  • *
  • Posts: 186
  • 34.502327, -116.971697
  • Liked: 105
  • Likes Given: 125
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1705 on: 06/21/2014 05:46 PM »
Thanks SwissCheese, I made the modifications as you suggested.

The i2p video has some black streaking after frame 161. This is due to error propagation. Frame 160 has always had a black dot in the ocean. Since this block is carried forward it gets duplicated each frame thereafter.

The next YouTube upload should have the i2p modifications.

Offline SwissCheese

  • Full Member
  • *
  • Posts: 164
  • Liked: 249
  • Likes Given: 92
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1706 on: 06/21/2014 08:14 PM »
Thanks SwissCheese, I made the modifications as you suggested.

The i2p video has some black streaking after frame 161. This is due to error propagation. Frame 160 has always had a black dot in the ocean. Since this block is carried forward it gets duplicated each frame thereafter.

The next YouTube upload should have the i2p modifications.

Could you also add the additionnal frames (credits...) to the YouTube video?

Online saliva_sweet

  • Full Member
  • ****
  • Posts: 555
  • Liked: 424
  • Likes Given: 1345
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1707 on: 06/21/2014 09:31 PM »
Replacements for frames 201 and 221 in i2p.zip. Fixed vop_quant. Difference is subtle for 221, not so subtle for 201.

Online Chris Bergin

Ooo, someone please message me if we are close to a new Youtube video! :)

Offline SwissCheese

  • Full Member
  • *
  • Posts: 164
  • Liked: 249
  • Likes Given: 92
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1709 on: 06/21/2014 09:44 PM »
Hey people, thanks for all the good work.

I just noticed that the latest You tube video on Twitter is two days old, Thursday 2014/6/19 13:02:58 GMT.

Is that intentional?

Just to explain a bit: we more or less finished repairing (aligning) the ("difference") p-frames, that's why nothing happened to the YouTube video.

What we are trying to do now is filling the holes of the ("reference") i-frames with the information of the previous frames. For this, Saliva_Sweet was able to convert the i-frames into p-frames.

Now we are trying to fill these holes and tune the color/brightness of the images to have smooth transitions. Most of them are done now, but a few still have issues.

Wronkiew will soon change the automatic video generation to use these frames.

Online saliva_sweet

  • Full Member
  • ****
  • Posts: 555
  • Liked: 424
  • Likes Given: 1345
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1710 on: 06/21/2014 10:00 PM »
Replacement for frame 81 in frames4_5 in i2p zip. Fixes most conversion issues. There's a small problem at the bottom right corner I don't know the cause of.


Offline wronkiew

  • Full Member
  • *
  • Posts: 186
  • 34.502327, -116.971697
  • Liked: 105
  • Likes Given: 125
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1711 on: 06/22/2014 07:09 AM »
Could you also add the additionnal frames (credits...) to the YouTube video?

Done. I also set the generator to output in 30i. Your suggestion to increase the output quality parameter helped there too.

Frames 261-280 are messed up because the new i2p mmb assumes that the segment starts at 221, but the generator is starting at 241. I think in order to fix this I am going to have to modify it to decode everything at once instead of splitting it up into segments. That's going to be a project for tomorrow.

SwissCheese, if you get a chance can you split the credits png up into two pages? It'd be easier to read that way.

Offline Quialiss

  • Member
  • Posts: 75
  • Liked: 82
  • Likes Given: 29
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1712 on: 06/22/2014 09:08 AM »
I've made a few small changes to some p frames in parts 5, 8 and 11.  A couple gray spots, the dark spot on the wave front and a bit more data in one frame that falls within the gap used in the next i frame. 

The dark spot on the wave is curious.  The macroblocks in all the wavefronts are quite large, 100-500 compared to the <50 average for most of the ocean.  I figured there was an error in one of the blocks and blanked out the first part where it appeared darker than it should, but there are either subsequent errors (3 frames total needed touching up to get rid of it) or I've chosen the wrong place to make corrections.


I'm going to try and get ffmpeg running locally again tomorrow to continue work on the i2p frames... last time I needed to use it it was throwing no errors, but outputting frames that were slightly different from the online editor.  Less than ideal.   ;D

Online saliva_sweet

  • Full Member
  • ****
  • Posts: 555
  • Liked: 424
  • Likes Given: 1345
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1713 on: 06/22/2014 09:52 AM »
Frames 261-280 are messed up because the new i2p mmb assumes that the segment starts at 221, but the generator is starting at 241. I think in order to fix this I am going to have to modify it to decode everything at once instead of splitting it up into segments. That's going to be a project for tomorrow.

Really liking the latest youtube video. But we do need to build it in a single piece at this point. There are a couple of instances where data gets passed through several i frames. I will go ahead and try to fix the chroma in frame 241 and then, I think, I'm inclined to call it done and my part in this will be played.

edit: Wait a second.. Has someone done it already?
« Last Edit: 06/22/2014 10:18 AM by saliva_sweet »

Offline SwissCheese

  • Full Member
  • *
  • Posts: 164
  • Liked: 249
  • Likes Given: 92
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1714 on: 06/22/2014 11:11 AM »
Frames 261-280 are messed up because the new i2p mmb assumes that the segment starts at 221, but the generator is starting at 241. I think in order to fix this I am going to have to modify it to decode everything at once instead of splitting it up into segments. That's going to be a project for tomorrow.

Really liking the latest youtube video. But we do need to build it in a single piece at this point. There are a couple of instances where data gets passed through several i frames. I will go ahead and try to fix the chroma in frame 241 and then, I think, I'm inclined to call it done and my part in this will be played.

edit: Wait a second.. Has someone done it already?

Could you check again the conversion of frame 61? I put on the spreadsheet a new intermediary set of mmb that looks quite good on the online editor but cannot manage to convert it correctly. Data seem to be missing/wrong at 41:3 and the bottom (18:28).
« Last Edit: 06/22/2014 11:31 AM by SwissCheese »

Online saliva_sweet

  • Full Member
  • ****
  • Posts: 555
  • Liked: 424
  • Likes Given: 1345
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1715 on: 06/22/2014 12:23 PM »
Could you check again the conversion of frame 61? I put on the spreadsheet a new intermediary set of mmb that looks quite good on the online editor but cannot manage to convert it correctly. Data seem to be missing/wrong at 41:3 and the bottom (18:28).

Thanks for pointing this out. There were major issues. Here's a new mpg4-img, mmb based on your intermediary and mmbgen file for the mmb porting tool.

Offline SwissCheese

  • Full Member
  • *
  • Posts: 164
  • Liked: 249
  • Likes Given: 92
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1716 on: 06/22/2014 03:17 PM »
SwissCheese, if you get a chance can you split the credits png up into two pages? It'd be easier to read that way.

It's in the attachment! (also made the last frame larger)

Frames 261-280 are messed up because the new i2p mmb assumes that the segment starts at 221, but the generator is starting at 241. I think in order to fix this I am going to have to modify it to decode everything at once instead of splitting it up into segments. That's going to be a project for tomorrow.

Yes that's probably the best way to do it, to do everything at once.

And once it is done, I think it is enough for me ;)
« Last Edit: 06/22/2014 03:18 PM by SwissCheese »

Online saliva_sweet

  • Full Member
  • ****
  • Posts: 555
  • Liked: 424
  • Likes Given: 1345
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1717 on: 06/22/2014 04:33 PM »
Do we have a consensus? Once wronkview can build the vid in one piece and uploads to youtube we call it finished and champagnes can be uncorked?
« Last Edit: 06/22/2014 04:34 PM by saliva_sweet »

Offline Roy_H

  • Full Member
  • ****
  • Posts: 918
  • Liked: 274
  • Likes Given: 1507
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1718 on: 06/22/2014 05:43 PM »
Sounds like you guys are almost finished. You are to be well congratulated for your stupendous effort! I trust Chris will post a new article with the finished version to highlight your amazing work.
"If we don't achieve re-usability, I will consider SpaceX to be a failure." - Elon Musk

Offline CraigLieb

  • Full Member
  • ****
  • Posts: 859
  • Dallas Fort Worth
  • Liked: 779
  • Likes Given: 974
Re: SpaceX Falcon 9 v1.1 CRS-3 Splashdown Video Repair Task Thread
« Reply #1719 on: 06/22/2014 05:48 PM »
Small suggestion to have the credit slides stay on the video a little longer so folks have a time to read it. Especially the one with the links.
Colonize Mars!

Tags: