Support forum for DG Tools

i would like a 10/12/16 bit HDR->SDR...

More seriously, i also agree with gonca on the part "you choose" if finaly one feature request is not more asked than another.
My request was more for curiosity to see what it gives, than a real immediate needs.

Also, i'm sorry, it seems i've badly asked my question.
You choose one method among severals for HDR -> SDR.
Would you share the reason why you choose it ? Unless you've already done that in one of the posts of this 20 pages thread ?

I tested with the available Vapoursynth solutions and liked Reinhard best. It is also easy to implement and has some good literature representation. I have to start with something; that does not rule out adding other methods later. As there is no current Avisynth solution it would be good to get something usable out there.

As I mentioned earlier I am an incremental rather than a big-bang kind of person.

OK, gonna work on high-bit depth support for the utility filters.

admin wrote: ↑

Mon Mar 12, 2018 11:39 am

10/12/16-bit support for my utility filters

This, at least 16 bit, as AviSynth+ 16 bit chain is almost complete. 10 bit as a necessary evil as h265 uses it in Main10 profile.

admin wrote: ↑

Mon Mar 12, 2018 11:39 am

CUDA solution for HDR->SDR. I would probably implement Reinhard tone mapping.

CUDA solution for HDR->SDR and possibly different dithering types even for bit depth decrease (at least none, ordered and floyd or bayer.

admin wrote: ↑

Mon Mar 12, 2018 11:39 am

Something else (please specify).

- High quality resizing (neural or any other hw possible thing)
- Temporal denoising (+ spatial, already existing)
- Make everything work as mtmode 2 in AVS+

Note sure if this of any use or interest https://developers.google.com/media/vp9/hdr-encoding/
under the heading

Convert SDR to HDR with a look-up table (LUT)

Converting HDR to SDR requires the transcoder to understand how to map brightness levels and colors to the smaller BT.709 color space and non-HDR range of brightness. For best results, an LUT describing how to perform that mapping -- usually specific to the content -- can be used. Examples C and D show how to use a LUT. We used bt2020_to_bt709_example.cube https://storage.googleapis.com/media.we ... ample.cube

Note: The LUT above is provided as an example of how to use an LUT with FFMpeg and is not optimized for the content in the example. LUTs are nearly always optimized for specific content.

Interesting, thanks!

Guest 2 wrote: ↑

Thu Mar 15, 2018 12:35 am

- Make everything work as mtmode 2 in AVS+

Sure, why not?

admin wrote: ↑

Sun Mar 18, 2018 3:04 pm

Sure, why not?

Would you please add the duration HH:MM:SS on the right of resolution and fps? And yes, when setting start and end of project range, the duration of that too...

Add them where? Humor me, I'm getting on in years.

Guest 2 wrote: ↑

Thu Mar 15, 2018 12:35 am

- Make everything work as mtmode 2 in AVS+

MT_NICE_FILTER will be better I think
As I understood from explanations and filter examples, programmer must care about memory allocations at getframe requests, not at the filter constructor. Mt mode 2 will care about safe memory allocation if filter allocates memory in the constructor, but it allocate, say, 4 times more memory with Prefetch(4). Imagine my double CPUs Xeon 2665 server running with 32 threads, so Prefetch(32) will eat a lot of RAM.

admin wrote: ↑

Wed Mar 21, 2018 12:36 pm

Add them where? Humor me, I'm getting on in years.

I think you misread. Where can you see duration?

OK, but maybe a status bar or something, not the window title. Lemme think...

admin wrote: ↑

Wed Mar 21, 2018 2:41 pm

OK, but maybe a status bar or something, not the window title. Lemme think...

Let's stay minimal. There is lot of space in the title bar, even when opening half res video.

While we are talking about interface. It would be nice to have something like this



in the status bar, while demuxing and saving project are running, i.e. visual feedback of progress.

You're a tough taskmaster. I'll see what I can do. How about a multiline title window?

admin wrote: ↑

Wed Mar 21, 2018 3:38 pm

You're a tough taskmaster. I'll see what I can do. How about a multiline title window?

Are you talking about duration? Never saw a multiline title window in my life. How is it possible?

Anyway, as most of us are using 16:9 screens, better to expand in lenght than in height.

Owner draw, of course. But I was being facetious.

There's already a cursor that runs along the timeline.

admin wrote: ↑

Wed Mar 21, 2018 6:15 pm

There's already a cursor that runs along the timeline.

I know but most of times I run it minimized. Anyway, no problem

I can't show the duration until after the project is saved. The entire stream must be parsed to count the frames. I do not use heuristics based on timestamps or anything like that, and anyway, elementary streams have no timestamps. So I could show the duration in the title bar after the project is saved or after a DGI file is loaded. It would be better, though, to show it on the Info dialog so that it gets placed in the log file. Showing the project range times would also be difficult for the same reason.

Is any of this any good to you?

admin wrote: ↑

Thu Mar 22, 2018 7:32 am

Is any of this any good to you?

For MKV too you can't get duration info?

Sure but I want to have everything handled the same. It's one of my primary design goals!

Hi.

A lot of Work seems to have gone into demuxing MKV files with subtitles and everything.

Would the same thing be possible for M2TS files, or when opening an MPLS playlist?

Right now when I open an MPLS file and choose the demux menu, I only get the option for video and audio demuxing. No subs.

And the naming scheme of the files also seems to be limited. In the demux menu the lanuguage of the audio tracks are written. But after demux that info is gone, and you are left with 1100, 1101, 1102, 1103 and so on. You then have to have the demux menu open and rename the files manually.

If I open the MPLS in mkvmerge, and create a MKV file, and then load that MKV file into dgindex and choose the demux menu, then I get the option for video, audio, subs and chapters. And maybe even more. and the naming of files after demux, also contains the language of the track in the name.

Just seems like going accross the river to get some Water, to first have to create a MKV file, with the soul purpose of demuxing that, rather then using the MPLS file in dgindex directly.

Thanks in advance, and thanks for a great piece of software.

renols

Sure, renols, the enhancements can be done for transport and program streams, and I have said that I will do it at some point. You're going to have to be a bit patient as I have a lot on my plate right now.

I can add the language to the demuxed filenames fairly easily. I'll do it but of course MEGUI users will hate me because the audio files won't be added to the main screen (until Zathor revises MEGUI).

I am an incremental not a big bang kind of guy.

admin wrote: ↑

Thu Mar 22, 2018 7:21 pm

It's one of my primary design goals!

And putting a ~(approximately) in front of duration and give correct one once demuxed?

admin wrote: ↑

Fri Mar 23, 2018 3:06 am

Sure, renols, the enhancements can be done for transport and program streams, and I have said that I will do it at some point. You're going to have to be a bit patient as I have a lot on my plate right now.

I am an incremental not a big bang kind of guy.

No problem. I didn't see that you mentioned that the enhancements would be addapted for other streams.

I will put on my "be patient" cap

renols