decoding frame

[cheyrn]

I almost always go through this routine when using virtualdub to edit:

* Get close to the area
* Go frame by frame until I get to the target frame
* Wait ~5 seconds if it says decoding frame
* Set selection start or end
* Jump back and forth until the frame is decoded

I am pretty sure that there are cases where it will decode after I have switched away and switched back where it would not say it was finished decoding if I did not jump back and forth.

Usually I have to repeat that process many times as I get closer to the target frame. It is particularly annoying at the moment because I am repeatedly trying to find the single frame in which someone has blinked. It becomes very difficult to tell if I am looking at the decoded frame or if it is stuck.

So, my question is: If you think the above makes sense, do you know of a quicker way to do this.

I think it also works to copy the source frame to the clipboard and then paste somewhere

[GMJCZP]

Hello cheyrn:

I use VirtualDubMod to preview, it allows you to F5 to refresh the image.
Now if you use VirtualDub and notes that costs you detect frame is possibly're using a complex Avisynth script that generates output not always uniform, maybe that's the reason.

One more thing, if you want to extract a frame for editing can take a look here.

[poisondeathray]

What type of source files ?

How are you opening it in vdub ? (what method? what decoder? )

You can "lose your place" (frames get mixed up) when scrubbing in vdub if you're not using a frame accurate method

Why is it taking 5sec to decode a frame ? What is the context ? Heavy filters ?

[StainlessS]

Perhaps only 1 key frame per gazillions frames is the reason (eg DivX, XVid),
try jump to next key frame and see distance between current and next key frames (a couple of times).
I would just render to lossless (and re-load that) and do away with all of the waiting.

[cheyrn]

Holy poop. Rendering to uncompressed and reloading has completely eliminated the problem for the first video.

The clips all have their speed heavily altered using avisynth (I am just starting to switch to vapoursynth). The speed never changes less frequently than once per second. It's sort of animation using video. But, the seeking problem is from the rendered video not the script.

The clip I looked at had a key frame after 300 frames then another ~150. Should I force key frames more frequently than that?

The source files are originally from Fraps. But, I am not usually editing the originally compressed video. I am opening them using file open with no other options. Should I use re-derive key frames on the files from Fraps?

Thanks for the frame replacing function.

[StainlessS]

DivX/XVid and other codecs that use a 'KeyFrame' are NOT intended to be edited again, if seeking backwards only 1 frame then will have to
locate previous keyframe then seek forward to requested frame. If using any kind of temporal processing in avisynth, then will need to repeat
that execise for every frame accessed that is prior to the current one.
I dont know much about Fraps except that I think it is some kind of game cap thing, but I habitually use ffmpeg to convert a clip to lossless
AVI (either HuffYUV or UT_Video) and do away with all of the problems associated with opening a source clip and finding best settings/sourcefilter.
It is usually quicker just to let ffmpeg do the aggravating bit.
Here a bat file to convert whatever to avi:

setlocal

REM Where to Find ffmpeg
set FFMPEG="C:\BIN\ffmpeg.exe"

REM Where to get input file, No terminating Backslash, "." = current directory
set INDIR="."

REM Where to place output file, No terminating Backslash.
set OUTDIR="D:\AVS\AVI"

REM Can add to below *.xxx style extensions for source file.
FOR %%A IN (*.wmv *.mpg *.avi *.flv *.mov *.mp4 *.m4v *.RAM *.RM *.mkv *.TS *.y4m *.yuv) DO (
REM uncomment one of below lines, 1st UT_Video codec, 2nd HuffYUV
      %FFMPEG% -i "%INDIR%\%%A" -vcodec utvideo -acodec pcm_s16le "%OUTDIR%\%%~nxA.AVI"
REM   %FFMPEG% -i "%INDIR%\%%A" -vcodec huffyuv -acodec pcm_s16le "%OUTDIR%\%%~nxA.AVI"

)

Pause

As above, save as convert.bat (or whatever) and copy source clips to same directory, will output to "D:\AVS\AVI".

You do occasionally find a clip encoded with only a single keyframe at frame 0, you would really have to wait a bit then.

EDIT:

re-derive key frames

I dont think would be any point, would just find and re-use same key frames, I think that is really only for when the clip
has been badly rendered or some kind of problem occurred whilst writing the AVI index (eg CamStudio sometimes writes
files with bad AVI index [or should I perhaps say 'sometimes writes files with good AVI index']).

EDIT: The only other source filter I habitually use is Mpeg2Source from DGindex which is pretty much always reliable.

[cheyrn]

Thank you. So I can create a 1GB lossless file instead of a 32GB uncompessed file and still have no seeking problems. That is certainly helpful.

I didn't try huffyuv because it says on videohelp that it's no longer developed and the link to the site is broken.

For ut_video I installed 14.2.1 from:

http://umezawa.dyndns.info/archive/utvideo/

For one clip, a problem with audio and video playing at different rates is fixed now going from avs -> utvideo AVI -> h264 mp4.

Maybe some of the problems I have been having with video and audio playing at different rates are a result of changing the speed of xvid encoded video.

[StainlessS]

If you have pretty much any codec suite or fdshow then HuffYUV will be in it, however if not and
you still want to try then here is 2.1.1 (dont use 2.2.0, bugged)
http://www.mediafire.com/download/i5...ffYUV+2-1-1.7z
Also includes multiThreaded version, and CCESP patch version (I cant remember what that is patching), and source.

[GMJCZP]

If you care about the size of lossless file (like me) you can try Lagarith (search in VideoHelp).

It seems, however, that UT is destined to become the leader in lossless process.