That is understood. However, I think that there should be the option to tell “auto” mode to put more emphasis on some of the parameters. For example: I think Proteus on Auto for frame-by-frame work is a good idea, but it would prefer that it should emphasize Revert Compression a bit more or Sharpen a bit more aggressively.
Proteus Auto is a very good idea, but it is very slow. I believe that being able to tell auto to put more emphasis on certain settings would help justify the loss of performance
What you’re describing is LITERALLY how it works. Set it to auto, then turn up just Revert Compression, and it will be turned up by that amount for the duration of your run.
I just ran a short 30 sec clip with no upscaling only Proteus processing and only with VEAI 3.0b. The first test was Proteus in Auto mode and that ran at 5.2FPS. Than I ran the same test but with a Proteus Estimated values, no auto. That ran at 5.3FPS. I would not characterize Proteus Auto as being very slow compared to regular Proteus. What is slow is 3.0 compared to 2.6.4 but that is known at this point.
I didn’t use any of them. All I did was try it on a 4k video which was already an H265 Main 10 by only using Slow Motion at 100% (tried both Chronos and Apollo) and Convert FPS at double the original?
It is also worth mentioning that the output quality of Proteus in 3.0 Alpha is much lower than that of VeA 2.0. The VOB file uploaded last time can reproduce this problem. Proteus manual Settings are: (from bottom to top, turn off prediction, pure manual) >> -100 2 0 2 0 1
[
-100 anti-alise
2 dehalo
0 reduce noise
2 sharpen
0 recover details
1 recover compression
]
The resolution is 6x. The main issues are picture floating, jelly feeling, mosaics, streaks, tearing, etc
It all depends on your source video and how compressed it is. For my material, I often find ‘revert compression’ auto setting is too high. In my case, the source videos don’t have many compression artifacts, so by turning ‘revert compression’ down, I retain more detail.
There is a speculation that it has to do with decoding, and the encoding method, I don’t know whether it is hardware decoding, such as Intel or Nvidia graphics card, low resolution no matter how high the bit rate is set, the final result will be Mosaic, just like the RMVB format video of ten or twenty years ago after being compressed many times. At least the GTX1660 can be trusted to encode, but it should be said that the CPU decodes and encodes better, except on top of the line graphics cards, and even then it’s partly for streaming and gaming, including AV1. It’s true that 3.0 wasn’t as good as 2.0, and when I tested it in 3.0B, the results were still poor. Or because of the lack of CRF/CQ to control quality and code rate, even if the code rate is set very high, even with ProRESS444xQ there is still Mosaic and confusion. I don’t know why, but it didn’t work out.
OK, I would like to say two things that I like about V3: 1) it’s ability to process up to 3 files at once and 2) copy/paste settings. And while VEAI is busy doing that, I can still do something on my PC. The direction VEAI is going is awesome!
As it’s now in Beta, does that mean version 3 is now feature complete?
A few comments:
Overall, I think the new interface is a step in the right direction. It feels like there’s still a lot to do though:
For me, and I realise this is purely a personal preference, the side-by-side view for preview is the worse of the views in 2.x. The view that I use almost exclusively is the single view - I find it far more valuable for comparisons. Will we be getting this back? I find side-by-side view to be of minimal value.
The comparison view from 2.x was a nice idea - but one I’d love would be to generate a bunch of previews, select them then use single view to flip between each, giving a very direct way to compare them.
A potentially killer feature that’s been more than once would be the ability to run scene detection on a video, then assign different enhancement options to different scenes - all to be run in the same batch.
A lot of the controls - playback / step / crop / zoom - seem very small and fiddly (as opposed to the huge and possibly oversized filter controls) - I find them quite tricky to use
Is the grain option going to be brought back? I don’t normally advocate features that aren’t strictly upscale / enhancement related, but that feature was very handy at times.
the real time preview must be implanted as even in this first beta, this feature is not working yet, so as this feature is totally missing and is planned, there is a high chance that it will be the same for the one frame preview !
I’ve been saying this and people keep saying it’s not true. It’s the same for previews. You can’t delete a preview even if you “remove” it in the software until you literally close the software. VEAI is not releasing file handles appropriately.
We absolutely should not have moved from alpha to beta yet, not by any standard form of product development. Nothing changed that made it “beta ready”…it just seems like it was an arbitrary decision.
