We have an alpha for v3.3 ready to test. The alpha has significant changes in various aspects of the application, so please let us know if you find any issue.
I don’t know if this includes output paths, but if they do, then that’s going to dramatically shorten our SSD’s lives.
If so, then might I add that there’s a lot of us who can’t afford to shorten the lives of our SSDs with a crap-tonne of temporary and massively heavy writes, some up to more than 5-10+ Gb per lossless video.
I hope I’m wrong, as I really want to test this alpha out.
NOTE: I may have to plan on purchasing an HDD for an alt startup system OS for this very reason…
EDIT: It takes me a while before I can install these alphas/betas due to having to use windows install/uninstall troubleshooter as I cannot install/uninstall these programs due to me having that cursed “deleting shortcuts” bug that permanently hangs the process, but yeah, I’m glad that this alpha doesn’t include output folders with the output settings issue!
As for the live-action, the blur seems to be large.
The difference between blurred and sharp areas is so large that for a moment it seems as if the image has become clearer, but in reality, the detail in the background area is greatly reduced. In addition, there is a large amount of color alteration.
What you should see in the sample is the loss of detail in the snow and the alteration of the color of the window, in contrast to the sharpness of the outline of the person.
As usual, the animation has bad afterimages. The intensity of noise reduction on the time axis is wrong.
I can understand if the noise reduction is 100 (CLI:noise=1), but this situation is clearly wrong at 10. Improvement is needed.
The test was run in the CLI, with REVERT COMPRESSION 20 (compression=0.2), RECOVER DETAILS 20 (details=0.2), REDUCE NOISE 10 (noise=0.1), and all others are set to 0.
Can confirm, plus the preview has the exporting dialogue with the preview, this happened with the second preview, and not the first curiously enough.
The major improvement, however, is that *the results are similar to a 2-4k res video even when the input is 244x432! This is some serious tech as it successfully reconstructed the face!
*NOTE: When the 244p vid is upscaled to 1080p, that is.
Edit: Turns out I did my testing wrong. The new Apollo Fast appears to fix the issue I reported. Thank you.
The new Apollo Fast model continues to show block artifacts on macOS similar to what was reported here for the original Apollo fast: Apollo Fast on macOS shows a gird artifact
Don’t see a way to choose between NVIDIA and AMD x264/x265 encoder. It’s important, because the x265 encoder of the AMD iGPU has higher quality than the NVIDIA GPU encoder. So, I use the NVIDIA 4090 for TVAI processing but the AMD 7950X iGPU for x265 encoding.
1/ Export as is don’t work
2/ Proteus V4 is very slower in exportation video than last bêta (Prores and H265, don’t try with h264)
3/ The preview is still not in line with the two, to immediately fix this problem which has been there for months
4/ The audio export with PCM don’t work whit h265 and h264 (it’s working with Prores)
5/ Why is nothing in encoder output in preference? It’s empty
Maybe something is wrong on my end, but I can not get the new Apollo Fast model to work on my Windows machine.
Information:
OS: Windows 11
GPU used for processing: RTX 4090, driver 532.03
I tried looking at the logs but I couldn’t find anything useful. But I have uploaded a copy to the dropbox. It’s called Alaska - Issues with Apollo Fast on Windows.tzlog
In a test video sent to me, which is massively compressed, the Proteus model included in this Alpha is massively better than Artemis Low.
It looks really good.
In the Artemis version everything looks very organic and morphed, in the Proteus version it looks almost normal or as normal as this compressed video can be.
Can Topaz Use Ai to pre Dect Blocking?
Video Can be Transcoded in to a Mid Stage File With 8 times res and Rec 2020 for Processing Than process the Mid Stage File for output.
Also outputions to input Projects From Final Cut Pro,Adobe,Magic ,Divinchi Resolve And Audio Version Projects. And Output Video With Project as well. Divinchi Resolve 18.5 Feature 4
