the whole video output engine from VEAI, is on the way to be rewritten completly for the next beta , from what the dev told, doing this will fix lot of things, more speed too etc. we’ll see.
how much speed you had with the 1080, and how much today with the 3060 and on which model ?
Hopefully, it will deliver an interrupt and restart of an ongoing upscaling Processes.
Nobody (at least not me) has a Pc dedicated exclusively to VEAI, and system restarts can/will be necessary while VEAI has notyet finished.
A backup feature “in the process” would be really really helful. but i doub’t it will be possible to do with a video output, it will work certainly with pictures, but it would be something very nice already !
I’m not noticing any speed difference with this update. Is it only optimised for M1? Should I be seeing it process faster with this update?
.45sec per frame on Artemis LQ 100%, using max VRAM slider to process clip
on
iMac
10-Core Intel Core i9
3.6 GHz
32 GB RAM
16 GB VRAM on AMD Radeon Pro 5700 XT
It is common practice for professional software to allow backup at customer predefined intervals.
I would be happy to have that with VEAI e. g. every 60 minutes when ETA predict 2 days processing. Such an interrupt of the process would be a negligible stretch of total processing time, and in order to save space any backup (x+1) should override backup (x).
2.6 Prores incorrectly uses double the bitrate resulting in massive unnecessary files… I’ve tried going back to 2.4 but in doing so the latest models no longer work stating I don’t have enough VRAM…
Is there any way to get around this, I really want to use the new models in 2.6 but cannot justify doubling the file size.
I have an M1 Mac Mini. Bought Giga / Sharpen / Denoise, but really wanted the expensive video tool. Just bought it and am stoked to see you have native M1 support which is really important for video processing!
Gonna use it on drone footage and upscaling political memes and conspiracy archives so that my material is nice and sharp compared to that 480p bitchute plebs.
I have an Nvidia p5000 as a graphic card but it is not used by the software. only the CPU works. I have selected the GPU as default but nothing happens. drivers are up to date!
There’s at least on the RTX3090 potential to be 4X faster than it is now. To see that a M1 chip on a notebook almost equals a 3090 in performance is such a waste.
It seems like VEAI relies too much on CPU even with GPU acceleration and this must be the bottleneck from the current engine. Lets’ hope for a future engine that will exclusively rely on GPU.
I noticed that a 1795 frame 30fps video doesn’t have a real-time ETA displayed at 0.24 sec./frame when using proteus v3… although it does have the ETA results with a post-processed file, is this by design?
On M1 Macbook Air, ProRes HQ output is no longer able to be imported into Adobe Premiere Pro. It says unsupported compression type. Tried installing ProRes RAW codecs, still not working.
v2.6 seems to be unstable for me compared to previous one I had installed (Did it jump from v2.4 to v2.6 or am I imagining things?). Trying to do a 1080P->4K Proteus upscale and keeps crashing about 30-60 minutes in.
First time:
Faulting application name: Topaz Video Enhance AI.exe, version: 2.6.0.0, time stamp: 0x618ae4fb
Faulting module name: onnxruntime.dll, version: 1.9.21.921, time stamp: 0x614a8837
Exception code: 0xc0000005
Fault offset: 0x00000000001243dd
Faulting process ID: 0x1e718
Faulting application start time: 0x01d7da23c8603f35
Faulting application path: C:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI\Topaz Video Enhance AI.exe
Faulting module path: C:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI\onnxruntime.dll
Report ID: 2d2adbb8-e50e-406d-8ef8-b3926fce7990
Faulting package full name:
Faulting package-relative application ID:
Second time:
Faulting application name: Topaz Video Enhance AI.exe, version: 2.6.0.0, time stamp: 0x618ae4fb
Faulting module name: opencv_world453.dll, version: 4.5.3.0, time stamp: 0x613a2889
Exception code: 0xc0000005
Fault offset: 0x0000000001bc9d24
Faulting process ID: 0x1bbe0
Faulting application start time: 0x01d7da50783c6503
Faulting application path: C:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI\Topaz Video Enhance AI.exe
Faulting module path: C:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI\opencv_world453.dll
Report ID: ce1ab778-8de6-4858-91e9-38c4b8c86228
Faulting package full name:
Faulting package-relative application ID:
EDIT: It did convert it the third time after 1h45m
I just ran in to the file permission issue with 2.6.0. I don’t know if this is a Topaz bug or Windows because what was causing the error for me. This happened on 2 separate computers that are almost identical. Both on Windows 11 also. What I found was my folder that had the files had a special character, “!” in the name. When I changed the folder name to something simple like “topaz” the error went away. This worked on both computers. I had never had the error in the past but I don’t normally have special characters in folder names.
I was reading on YouTube a person said proteus settings are locked they cant click on any of them they said even when reinstalling the program it still had the same problem
I had that problem before but just exited and started up agian and it worked what would be the fix for that if reinstalling didnt work
It also happens if the output is sent to a drive on another PC connected to the VEAI machine via RDP (which is the setup I currently use). Even if the drive is mapped with a drive letter such as Z: instead of using the network path VEAI will refuse to save to that path. I got the external PC’s drive to be accepted for one session by entering the drive path directly into the registry but that did not work in the next session (VEAI closed and reopened).
For the moment I work around this bug by sending the output to a simple local path on the VEAI machine and running a batch that periodically shoves the content of that path to the drive on the machine connected to the VEAI machine via RDP. But I’d rather not have that issue - the bug didn’t occur in 2.4 so it should be feasible for the devs to rectify what triggers the error in 2.6.