The color space issues appear to have been resolved with my test videos. The application also failed when any AI processing was enabled, and this has also been fixed. Thank you.
I have two GPUs installed. I find that processing fails immediately when setting “All GPUs”. Selecting one GPU or “Auto” works fine. The max memory setting is on 70%. What does Auto do by default?
Hola! beautiful interface, for color change, it’s solved for the deinterlacing, always the same, dione td and dione dv deinterlace very well but the problem is that the artifacts are clearly amplified which makes the video irretrievable. Then, for the preview, it always bugs, it does not play in loop at times, as soon as when I stop the process as soon as I see that the image changes in the preview.
Thank you for adding the clean uninstall capability for Windows, too.
@ida.topazlabs
It would be wonderful if someone could make a list of the areas/folders where temporary or obsolete information was stored in previous versions? I would love to be able to dig that all out and delete it. It’s not just a messy waste; eventually it will attract bugs.
I have 2 GPU also, and it fails on mine as well. This used to work very well on the earlier versions of the beta, and I can confirm that with both all GPU selected, I was getting about twice the throughput. Not always, but yes when i was doing enhancement and slow motion at the same time.
Hope they fix it soon. I did send an error report last week.
Thanks for sending the files and notes. Is the color shift issue occurring for all encoders or any specific encoders? We are aware of this issue with the 16-bit tiff and png sequence.
@nipun.nath there is still no list of models as it was in version 2.6.4 where you could choose, it would also be nice to put the old models. feel like they don’t read any of our posts to take notice, this is the third time I’ve written this.
I really like the new GUI. It feels great to use, is very responsive and looks good in my opinion. I also like the new icon.
Previously I had one video with colour issues, but it works now. And again no new issues either. Looking forward to the final 3.0 release.
Ok, been too busy since b9, back on it with b11. Nice improvements to UI appearance.
What I’ve noticed in first 20 mins (and I focus a lot on my large library of interlaced archival footage).
Launching b11 from the installation box resulted in invisible UI but app button on the task bar. Closed the button and launched from desktop icon, UI operates normally. Except “Unmaximize” button needs TWO clicks to unmaximize, as I reported for b9.
Deinterlace still not restored as a first-pass, can only choose it as an Enhancement, which means I can’t use enhancements. This is a show-stopper for final use.
When interlaced and using x2 Dione model, when Out is set to “Original”, output (Preview) is still x2; e.g. input is 25fps, Preview is 50fps. Can we PLEASE get a grip on this? It’s been a problem for numerous betas now.
Having done 4 Previews in a session and wanting to compare them, switching now sometimes results in the app hanging (endless twirler). Have to close and start again.
Preview of source (interlaced) still flicks like a mad thing, along with mad flicking of pointer on play bar. Preview clips still sometimes play repeatedly, sometimes only play once when you click the Pause then Play button. Been a problem for many betas now.
Question: How do we tell what order processing is done in? For example, is deinterlacing done before stabilization (as it should be), and in what order is upscaling done? I see that temporal interpolation has risen up the application list… does that mean it is now done before Effects? Can we choose the order so we can test whether different orders give better/worse results depending on the characteristics of our source footage? Do I ask too many questions?
Codecs with YUV colorspace are affected.
Please check the additional sent items that I tested with H.264 (NVIDIA) and H.265 (NVDIA).
I have verified the differences in color space in DaVinci Resolve and TMPGenc.
Processing time is disappointing. With the release before I could run 2 in parallel with 2x ffmpeg @ 25%. with this release 2xffmpeg @ maximal 14%. Processing time has increases from 100% to 180%. That is inacceptable!
Setting the input framerate still doesn’t update the scrubbing timeline to the correct time frame, so half the timeline is stuck on the last frame (same issue as last reported).
Related to above, trying to clip he sequence still has the right side go off the screen - though funny enough now the bar itself is actually going off the screen as well, which makes it more funny to look at:
Once you “clip” it by manually bringing back the right side, going back to main sequence the bar seems fixed.
Still missing frame selection.
The export locations and selections still don’t follow standard usage of the Explorer in windows, which makes using it very awkward.
Asking for a “name” to save the sequence, only for that name to be interpreted as a folder instead doesn’t happen anywhere else I am aware of and while I assume its intentional, feels like a bug and the program is broken. It also makes saving location painful when doing multiple sequence. This also results in cascading folders still
If I want an export to go to the same folder, why is that function impossible to achieve? If you try to select the same folder, it wants a name. If you give it the name of the same folder, it actually opens that folder then asks you for a name again - once again, this is abnormal behaviour for Explorer standard usage and feels really clumsy.
The above was handled beautifully by TVEAI for years - I really would like to know whom thought current output was a good idea, it wasn’t broken before. I also cannot understand how changing the output sequencing from something that makes sense, to something that makes no sense, is supposed to make the new UI easier to use. If you want this to work this way, so be it, but add an option in preferences to revert back to a normal expectation of saving in Explorer.
Clicking preview, and then clicking close preview does not stop the processing of that preview. When you are testing various models and grain etc, you often preview things over and over again. This means clicking that button a fair amount. However, instead of clicking stop preview and that being the end of it - I now have to stop preview, then close preview, then click a confirmation box that i want to stop the preview, tripling the user input for a simple task. Terrible UI behaviour.
Preview is STILL broken from last release. It will not preview the section you are trying to look at. Scrub to a point, click preview to see what it looks like and be beholden to a preview of a different part of the clip.
I see no fixed issues in this release for UI from issues previously reported. I don’t mind testing, and happy to keep testing, but when I see the same broken behaviours release after release with no fixes occurring, it seems like I waste a lot of time testing new releases with the exact same issues one after the other.
I think I will wait now until the fix notes actually indicate some fixes before testing a new release.