VEAI New Features Request

Thank you for your response, Mac; however, I believe multiple audio tracks and subtitles/cc are the future. Every podcast host, tutorial instructor, media personality sees an increase in traffic/engagement with closed captions/multiple languages of their top market(s).

Video sites like Vidyard allow multiple audio track uploads while YouTube allows additional audio tracks to be added. It’s only a matter of time before everyone follows suit with the 1-click upload. which is good for business.

Today, Youtube is murdering Hollywood with creators making content with DSLRs and Movie Maker, lol. “The Independent Content Creator” is the market every company the size of Topaz Labs should be marketing to intensely. Views for Videogame play-through videos are in the BILLIONS with 4K being in demand. Hello, VEAI 4K upscaling!

I want as few workarounds as possible and if OpenSource HandBrake has these features, why not VEAI?

And you are correct, VEAI is not meant to be a complete video editing suite. It is destined to be a complete post-production suite!

1 Like

Mac and axymeus,
if VEAI and the Topaz AI De-interlacer were released as plug-ins (Adobe/OFX) instead of a standalone software package, could development be easier or is that another can of worms?
Thank you.

To be really honest, if they were released as plugins, I fear Topaz might follow the stupid Neat Video route and release a different version with a different licence for each major video editor.
The issue with this is this does lock the customer to a specific editing software.
This, imho was a dumb marketing choice from ABSoft, and is the only reason I never bought the complete version of their plugin.
But if people at Topaz Labs were smart enough to release these as plugins for all editing software (in the sense 1 license allows 1 customer to freely use the plugin), I have strong belief they would sell many licences.
Thats actually what makes the standalone versions quite successful as it does not constraint the customer.

To give your question a more direct awnser, yes plugin versions might seriously help the workflow. But as every plugin, it would depend on their stability and their available options.

Last reply related to your previous reply about multiple audio tracks. I am perfectly aware and I totally agree with you about the future being with multiple audio tracks. What I meant is that for the moment, all we’d need is VEAI to be able to pass them all through without processing. That would be amazing. Currently it kindof destroys the single audio track in low quality, and that often forces us to save the original track and remux them. So if it did the same to multiple tracks, it would not help in any way :slight_smile:

2 Likes

Sorry, if this was posted elsewhere already and I overlooked it. What would be a very nice feature is frame interpolation via DAIN, CAIN, … The upsaling of VEAI is very impressive. It helped me rescuing some old footage taken on horrible mobile hardware. It had only 9 frames/sec and without frame interpolation you get a bigger, nicer video which still is some sort of stop-motion-video. Extracted and interpolated the frames and then used VEAI, but it was quite some hassle and I think, this would be feature which fits the idea of VEAI pretty well.
Thanks for reading

2 Likes

Please add ability to select input video aspect ratio. So far we are forced to make a copy in some container (like mkv) where we can set such aspect ratio.

2 Likes

I have been using the software for processing home videos and also a number of TV episodes.

So far I have been going through episodes of Star Trek Voyager and The High Chaparral amongst others.

I found that if I put a whole episode through the software as one file of 45 to 50 odd minutes the software would crash before completing the episode leaving an unusable MP4 file behind.

This would mean having to start it all over again.

My workaround has been to split episodes into chunks and then merge them together again afterwards.

This was a few versions back so the problem may have gone away with the latest release but because it takes so long to process an episode I didn’t want to have the risk of having to start again.

One of the enhancements I would welcome would be interim saves and resume points.

Creating a resume point every 30 minutes or so of processing would allow you to pick up where things left off in the event of a failure.

My other suggestion would be coping better with 4:3 material and for instance the ability for the HD (1920x1080) preset to crop to the actual image area.

1 Like

Interesting. I know you can resume an image sequence export and/or media download; but unaware about a .mp4/.mov. Would certainly welcome that, if possible.

Try exporting as an image sequence and remuxing/exporting final video with the original audio track.

Thank you for the input, Glenn.

Agreed. Gigapixel, Sharpen, Jpeg2Raw, Mask and Denoise AI all work as stand-alone and Photoshop plug-ins, so hopefully, Topaz Labs will follow suit with VEAI being a standalone and Adobe/OFX plug-in (Pro Version).

is there a way to know which Artemis model is better for my 4K daytime drone footage taken with Mavic Air 1? Artemis HQ9, Artemis MQ9, Artemis LQ9 or Artemis AA9? the description lacks any viable information how to decide? Doing previews in 4K is time consuming. I have quite a lot of videos I want to process in VEAI before using video editing software. Talking in tens of GB data.VEAI%20181%20Model%20Options

I would like to second the arguments for direct audio copy / passthru as the purpose of the software is to enhance the video, but not degrade the audio.

I also think an autosave feature is desperately needed as the conversion of video is a time-consuming process and losing time with failed conversions is time we can’t get back.

I also have not found a way to apply the same conversion settings to multiple videos (is there a way that I’m missing?) and so would like a feature to allow you to apply the same conversion settings to multiple videos as right now I have to make settings individually for each file.

1 Like

Quality is quite different from one source video to another. There are hundreds of variables, including type of camera used (film, tape, digital), SD or HD, clean or noisy, compressed or uncompressed, sharp or soft, interlaced or not, etc. Also, every person has different preferences of what they like. Some like a grainy film look, some like it soft and smooth, some like it very sharp, some not too sharp, and so on… well you get what I mean. So, if there was a specialized VEAI model for every type of purpose, there would need to be hundreds of them.

Topaz has trained a handful of models to try to accommodate various sources and preferences. I have been using VEAI for many months and I have come to the conclusion that the best way to decide on the best model is to run a preview of all of them and use my eyes to determine which results appeal to me. In my opinion, it is worth to spend some time to do that. I tend to gravitate to Artemis HQ and Gaia HQ. I usually keep away from Gaia CG and Theia. For SD sources, I often do a 100% Artemis HQ run to clean up the video, followed by Gaia HQ to upscale and bring out more detail. You might select different models than I, based on your own eyes.

1 Like

The ability to save videos in other formats at the point of being processed in VEAI would be good as having to turn thousands of TIFF files back into a video after the fact is time consuming. I encode in MagicYUV predominantly. Also, is it possible to scale non-uniformly as I work with a lot of non-square pixel videos like VHS and DVD and I have to scale to the correct ratio in Virtualdub afterwards.

For example, I have a file that I want to output to 1910 x 1080 exactly from a non-square source. but to achieve 1910 resolution, I have to uniformly scale it to 1910 x 1152 then scale the processed video to 1080 later on. I want to be able to tell VEAI to non-uniformly scale as it processes, thereby skipping the later process and maintaining absolute clarity.

1 Like

I have suggested once of VEAI having the ability of being a frameserver. Creating a postsigned uncompressed AVI video file that then could be loaded on many other programs to then encode/postprocess directly.
Or at least like VirtualDub, the possibility to use vfw codecs installed on the machine. I also encode all the masters videos from PNG or TIFF with the lossless codec MagicYUV.

And let’s go even further with a framerate converter at the same time of upscaling (DaVinci resolve has now a neural engine called speed warp that recreates in between frames with much less artifacts than the traditional optical flow methods).

Is it possible to download models separately from the main application?

Yes. Somewhat complicated, but a number of us do it with each version. I wrote instructions a couple of months ago that I just updated; it is titled Model Download Tutorial and you will find it here in the VEAI Discussion section.

Now, VEAI cannot make a video in1920x1080 pixels, from a 720x576 16:9 , or 720x480, etc etc.

Please add this option.

It can, if you check the box that says “crop to fill frame” or something. A 4:3 will become a 16:9 this way, but naturally you will lose some of the picture. If you’re talking about anamorphism on the other hand, you can simply set the apropriate DAR in the final video container to achieve what you want.

  1. Adding “Open with Video Enhance AI” to the Windows 10 Context menu (right mouse click button)
  2. Adding ability to open VEAI from other software, in example Fast Video Cataloger (External Video Editor) works with VirtualDub but not with VEAI.

VEAI%20Windows%20Context%20Menu Fast%20Video%20Cataloger%20Preferences

I’m wondering what the balance between spatial and temporal denoising are for the various algorithms.

I like how Theia allows us to adjust settings for sharpening and deblocking, but as far as reducing noise goes, I’d like to adjust settings based on spatial and temporal elements to denoising. Often times Artemis blurs too much when there’s enough grain and a fair amount of motion, but I really like its overall detail recovery - hopefully we can see a new setting for bias to temporal denoising over spatial.

So far, VEIA is doing a good job denoising, sometimes upscaling, sources already of a decent to high quality. I would love to see two new models emerge: one trained specifically against low bitrate videos to get rid of blocking and other artefacts introduced by high compression (as it stands now, these artefacts are simply upscaled alongside the rest of the video, no matter what model I choose); and one model trained specifically for deblending (I already proposed that but I thought it couldn’t hurt to have a trace of it in this feature request thread).