Apparently not a worthwhile question.
Iāve tested a few videos and Frame Interpolation past 10x does not work. it says Model loading and just stays there. Any one have this issue? im on rtx 3080
EDIT UDPATE: seems to be just apollo model doing this
I am running into all kinds of issues when trying to process VOB files. The main issue is with trimming where setting the āstartā and āendā are not very responsive (the position of the markers gets displayed a few seconds after moving them). Once I start exporting only one frame gets processed and it goes straight to 100%. I donāt see this issue with other input formats. The other issue is abrupt crashes as soon as the model is loaded and processing starts ( also only observed when using VOB files. )
Apparently asking a question specific to your use and not getting a reply from others triggers tantrums.
Hi, Iām using it with Windows 11 since the first version and Iām very satisfied with the results (at least mostly).
BUT, thereās one issue about the āpinned iconā of this software in the taskbar, that Iāve been facing since the beginning.
EVERYTIME I update it, the pinned icon becomes inaccessible from the taskbar, so I have to delete it, open the app again, and then pin it again. (not to count that I have to reposition the icon to the right place again inside the taskbar)
This never happened for me with any other app Iām using even after updating them.
Am I the only one having this issue or it is to be considered ānormal behaviorā?
(By the way, the same thing happens with Photo AI)
Thanks in advance
This AM doing The Italian Job 4k⦠Resizing with some black bars so it is perfectly 4k resolution, but the rest is just proteus, mostly reduce noise and deblur⦠getting about 2.0fps on my 3080 this AM.
Iāve encountered same message as well. Funny enough, I try to download same model later in the day, or rather during nighttime, and it worked with no error. I wonder if its related to server issues on the Topaz side, being not enough capacity to handle all requests. Just thinking out loud.
Hmm. I didnāt realize that. Strange. To me it happened during the ādownloading modelā stage⦠I think. But I didnāt pay too much attention on all the things it was happening in all the progress bars, so I canāt be certain if it happened during downloading stage or loading model stage. But Iām pretty sure it was not during processing stages. That is strange, indeed. Bug, probably. Iām just guessing again here, but I wonder if it happens during processing when programs detect a frame for which it would normally be trained on another model and then it stops because itās not āideal modelā.
I wonder if that is how it works since every time you load a new type of footage it analyzes it and then downloads the optimized model. What if during processing it found a frame that for some buggy reasons should be done with another type of optimized model and instead of ignoring the error, it stops and gives the generic error message. Iām not a programmer, so Iām just throwing ideas at the proverbial wall here.
More importantly, we need them to add support for the new AI cores in RDNA3!!
i have the same issue.
Im not sure if I remember correctly but back in version 2 Topaz dropped the installer they were using and had to write their own. Thus probably why the icon doesnāt get updated.
The way I deal with this is to unpin the icon just before I install the update. Once I get to the last screen of the installation I check the box to launch the app. Once launched I just pin the app again. That seems to be the most efficient method for now.
Thatās because they setup the shortcut weird. If you right click on the desktop shortcut and select āopen file locationā it opens the user desktop folder.
If you were to find the program in your āProgram Filesā folder and select āSend to Desktopā it will create a shortcut that will not become invalid after each update.
Thanks for that. ![]()
Hello! I really like your program, but there is one big problem that spoils the whole impression.
My system: Microsoft Windows 11 Pro, NVIDIA GeForce RTX 3060 (12 GB), Intel Core i7-12700KF, 4900 MHz, 32627 MB (DDR4 SDRAM).
Now I am converting (HD to 4k,120fps) or (UHD,4K to 120fps)
No matter what I convert to 4k, 60/120fps, close-ups, faces turn out well, but when the camera moves, in the background, on trees, rocks, skyscraper windows, gratings, grids, unbelievable noise appears that makes the video unusable to view. This is a catastrophe!!!
Summary: Your program cannot correctly convert meshes, cells, honeycombs, trees, grass, soil in the background when the camera moves. Can you do something about it? And do you know about the problem?
the preview screens are too small please make them use the space up
Iāve been bunkering down in 2.6.4 as of late, I was wondering how the quality of this newest build (3.1.2) compares to 2.6.4?
I noticed that on some releases that there was a āquality mismatchā between the 3.x.x versions and 2.6.4, has that changed with these new builds?
Guessing youāre using Proteus auto? Change it to manual and make sure the Anti-alias/Sharpen slider is at 0. Turn down Recover details and Sharpen too. You can have them higher than 0, but they start to introduce those patterns youāre seeing. (Sharpen more so than Recover Details.)
Does Topaz know about this? They might not.
Personally, I only upscale DVDs to HD, and I have found that without exception, Proteus Auto and Relative to Auto, always produce those patterns you mentioned.
No! 4k 24fps to 120 (60) fpsā¦Frame interpolation, Cronos fast
Oh okay. Iāve never see that from just Chronos, but with all the upscaling models except Gaia. What I have seen from Chronos are sections that jump around. Apollo produces much less of those, but at about one third the speed. I have also never had a 4k source, so I cannot know if it only happens at that scale.
Can this be:
When exporting a 13 second video with 2-pass stabilisation Topaz Video AI v3.1.2 shows me remaining time approx 3h on an Intel Core i5 12400 with 6 Cores???
I enabkle Strengh=50
Full-Frame
Rolling Shutter Correction=Yes
Reduce Jitter Motions=Yes
How can I speed up re-encoding?