I edited your comment to remove the sarcasm, in case you don’t know this is a user to user forum and, as such, the assistance comes from other users.
My reply was just to say that you will get the technical support you need direct from the Topaz Labs support.
Only the GTX 10xx and higher series are currently supported. I don’t know whether there are plans to expand that range.
OK, thanks, I guess I have to wait using the SW then.
Removes video compression artifacts and increases detail
automatically exports in h.263
I’d enjoy it if the initial enhanced video was exported in a high bitrate AVC or HEVC, both so that the recovered detail could be enjoyed and so that there’s a master to optimize from to reduce generation loss.
I’ve found a workaround to this. What I’ve done is export to PNG, then use Images To Video 4.0 set on max bitrate to mov (Which uses the H.264 codec) to make the video, and then I use MKVToolNix to merge the audio from the original video into the enhanced video. (I like to copy the original audio stream exactly instead of re-compressing it to prevent generation loss)
use the h263 file and reencode to HEVC with handbrake. 1 steps less since V 1.1.1 lets you maintain audio from source so you don’t need both MKVToolnix and Images to Video. Handbrake will reencode the video at your settings and keep audio from the file from VE AI. However Video Enhancer AI should automatically encode the video to HEVC in a best case scenario so we could get obth a better looking and reasonable size file.
I am trialing the product. However I noticed immediately that it does not appear to support Sony AVCHD video. Why is that? And is there any intention to do so? If I have to re-encode than this product would not be acceptable.
Where do I ask question about the actual features of the product, though. The support request option seems to be for problems with the software, not general questions?
I would like to know if the x200 and x400 enhancement options also denoise/deblock the file as standard or is the x100 denoise/deblock option the only setting that performs denoise/deblock? Where should I direct this question, please?
Handbrake compresses all files, so you will lose detail. It has no pass through option for conversion.
It is for technical support so direct your question there and @bowen.wang should be able to answer your specific question about:
VEAI files are not lossless anyways, so provided that you keep the bitrate high enough you’re not losing anymore detail that you would’ve already lost
I would be surprised if AVCHD becomes a properly supported format.
Unlike mp4 and similar containers AVCHD is a format with a folder structure with multiple files (a bit like a bluray). The video is stored in .mts files. So long as the video was filmed in a progressive field format, you could try opening a .mts file in VEAI to see if it works.
I have been using Video Enhance for the last 2 weeks. I have much feedback to give. I will try and be concise and clear with my points.
First is this software feels very much in development. Basic needed and expected features of any published software are missing.
Confirm about Exiting the program while it’s in the middle of an 100s hour long encode. Currently, it just exits losing all work.
Allow work recovery, when exited.
Allow work to be paused/saved/picked up where left off.
Recovery of work should be possible as the incomplete mp4 file is written too in chunks and on the drive. I have already lost 100s of hours of encode time from accidental closing of the encoder.
Now that the basics are covered, let’s move on to some bugs I have run into.
Most of the outputted files do not sync back to the sources audio. Output has different frame counts and frame rates. As the purpose of this software is to improve quality of clips and then edit them into a time line, this seems important.
Why is this advanced software outputting to DIVX h263? This just seems odd. I sent a example to a college and got laughed at for the DIVX video.
Seems a lot of performance is left on the table. Why isn’t the software using the Inference Hardware on the new RTX Nvidia graphics cards?
Also is there a way to make the long running encoder play nicer in the back ground? The machine running Video Enhance, even if very powerful is unable to be used for other tasks. As an example all my video encoding software while able to use 100% of 32 threads is an idle tasks and we can use that machine to play AAA games with no noticeable loss of performance. Currently while running Video Enhance this machine can’t even play back video.
All in all I think this software shows promise. Just needs some more development!
I’d like to know this as well, can you let us know when you get a response from tech support as suggested by AiDon (as opposed to both of us asking them the same question.)
Thanks
You didn’t specify what kind of hardware platform you’re running on. I minimize it to the Win10 taskbar and running in the background the performance degradation isn’t even noticeable, I can also play videos without issue. I’ll admit I’m running on an Intel 9900k with an RTX2070, but it’s only using about 30% CPU and 20% GPU which to me indicates it could run on a lesser CPU/GPU combo without issue.
I have tried it on an i7-8700k @5GHz, 48GB RAM, with 1080ti as well as an Ryzen 3700x, 32GB RAM with RTX 2060.
The task does default to Normal Priority in taskman. The software uses all of the GPUs RAM 5.8GB on the RTX 2060 and 10.7GB on the 1080ti.
I tried it on a laptop i7, 2060, and an Ryzen with 1050ti and gave up in both those cases. Too long too hard for the laptop and the 1050ti desktop estimated 2000 hours as well as not making any progress (wrote down frame number) while monitor was turned off.
It does not seem to denoise, as on grainy content it seems to just exaggerate the grain and create horrid moiré patterns. Didn’t seem to matter if I choose the HQ or LQ presets.
Both questions are still to be answered and have been already brought up from another users (me included).
For Codec, my own personal opinion is that they should go for H265 as it’s not only the most recent derivation of mpeg4 video, but’s also the codec created with UHD video in mind. However h264 would be accetable for legacy compatibility and seems to be the codec most users would want (from some samples here in the forums). The best would be to enable the user which codec he would want to use.
As for RTX tensor core use, from what I know of tensor core usage, they require FP16/INT8 usage, and VEAI image processing is build purely on FP32, so as of how it’s made right now, it’s impossible to use.
If they want any bigger adoption of their app, Topaz Labs need to make a FP16/INT8 module that can call for Tensor Cores and therefore get huge boots in performance with RTX cards (and possibly future AMD cards if they implement something like tensor cores, which I believe they will do, sometime in the future).
I can confirm to you that VEAI is possible using anything between 80 to 100% of your GPU while encoding a video. Windows’ task manager doesn’t do a good job at showing GPU usage. You should look into Hardware Monitor or other similar app, as they are more trustworthy in this. I too have a RTX 2070, and it usually runs at near 100% while encoding a video with VEAI, with less usage sometimes, but never below 80%. Your GPU fans are likely to be running full speed while using VEAI too
We just tried playing For the King, a casual adventure game, while the 1080ti was running Video Enhance. The game ran, but after we exited the game Video Enhance was no where to be found. We lost 50 hours of encode time.
