Problems with LAV Decoder/Splitter ripping thumbs from TS files using SUU Video Thumbnail Maker [newbie, be gentle;)]

Hi all, I am new to this forum and reasonably new to whole codec 'thing', so I hope you'll forgive the newbie questions  

I am a user of SUU Design's Video Thumbnail Maker. As the name suggests, it generates thumbnail pages from videos - generally it does a great job. The videos I am ripping, and the reason I installed the K-Lite Codec pack was to rip thumbnail pages (index pages) from lots of TS files which generally have these properties:

Format : MPEG-TS
Overall bit rate mode : Variable
Overall bit rate : 500 kb/s
Format: : H.264/AVC
Color space: : YUV
Chroma: : 4:2:0
Bit depth: : 8 bit
Scan type: : Progressive

Importantly, VTM allows the user to employ either Inbuilt Codecs or, when they have problems ripping, External Codecs. Since the Inbuilt Codecs are limited and error prone, the K-Lite Codec Pack was recommended to me.

I spotted the recommendation "We highly recommend to use the (default) LAV decoders" on the Download page, so I did some initial testing with K-Lite Codec pack (Mega) and quickly came to same conclusion - ie, that the LAV Decoder/Splitter was the best for TS files (such as those above), however .... after ripping hundreds of files, I noticed that files generally larger than about 100 MB failed to rip/thumb/decode - which I assume is the fault of the LAV Decoder. Since a more than 40% of my videos are 100 - 1000 MB, I am missing the ability to thumbnail a high proportion of my collection of videos.

I have a Ryzen 8-core PC with 16GB of RAM, and a 3 GB 1060 OC nVidia card, so horsepower isn't really an issue.

So, I have a few key questions at this point:

1. Are any useful logs stored by the LAV Decoder that might give some clues as to the cause of failure?
   
2. Are there any other setting tweaks I should I consider? (current LAV Codec settings and LAV Splitter settings )
   
3. Should I be preprocessing the TS files in some way to make them more stable? If so, what do you recommend?
   
4. Should I be considering an alternative Decoder for these files? If so, which?
   

Any and all thoughts welcome! I can provide samples if needed.

Many thanks in advance

The main difference between TS container and other file containers is that TS does not have an index for seeking. That means that seeking is slower because the splitter has to parse through the file to find to correct spot to seek to.

You can test playback with the GraphStudioNext tool from the pack. If it works in there it should work in your application as well if you do things right. No special settings are needed for playback of TS. It should work out-of-the-box. If you have any files that do not play, then send me a sample.

LAV does not have a logging option that you can enable. You should check the return values of the DirectShow functions that you call. That should reveal where the problem is. You might be calling a function that isn't supported by the splitter.

Thanks for that information and those tips.

Admin wrote:You can test playback with the GraphStudioNext tool from the pack. If it works in there it should work in your application as well if you do things right.

I've run a random test file of the same type (which also doesn't Decode properly to thumbs) and it appears to be using the correct LAV Splitter and LAV Decoder. Playback also works fine on the inbuilt player or Potplayer.

Admin wrote:If you have any files that do not play, then send me a sample.

I have sent you a random file of the same type which, whilst it does play, does not rip properly. Perhaps there's something you can see that problematic with the file that I cannot.

Exactly how are you extracting the thumbnail? I have no idea what you might be doing wrong if I don't know what you do.

For example IMediaDet and SampleGrabber do not support decoders that use VIDEOINFOHEADER2 structure.

Admin wrote:Exactly how are you extracting the thumbnail? I have no idea what you might be doing wrong if I don't know what you do.

As per the subject line, I am using SUU Design Video Thumbnail Maker to create thumbnail sheets from individual videos (not so much single thumbnails of each video).

Here's an example of the output:

Here is some example code to extract image from VMR9 video renderer:
https://stackoverflow.com/questions/11398758/best-directshow-way-to-capture-image-from-web-cam-preview-samplegrabber-is-dep/11400807

You will have to manually construct the DirectShow filter graph to use VMR9 renderer, because system default renderer is VMR7.

Admin wrote:Here is some example code to extract image from VMR9 video renderer:
https://stackoverflow.com/questions/11398758/best-directshow-way-to-capture-image-from-web-cam-preview-samplegrabber-is-dep/11400807

You will have to manually construct the DirectShow filter graph to use VMR9 renderer, because system default renderer is VMR7.

Unfortunately I neither have the skills nor the time to take on some coding - hence my choice to go with a prepackaged solution which integrates the great K-Lite codecs. Unfortunately, the codecs are failing in most circumstances where the file exceeds 100 MB (ish) - it's hard to know whether that's VTM or K-Lite. It appears we've hit a dead-end

Oh, I though you were the developer of the program. The bug in their software, so I can't be of much help.

If it only happens with TS files, then I suggest you remux the files into MKV container. You can do that with MkvToolNix tool. It only takes a few seconds per file. Benefit will also be that seeking will be fast.