Until recently, I was doing something like that, too. But recently, I discovered that you can get at least as good, usually better by running it through Proteus Auto at original size. You can even kick up denoise a bit more. Artemis high-quality won’t always take out enough noise and going to Artemis Normal loses detail.
How can I remove chroma noise with Proteus? Have tried lots of settings but it seems it sees chroma noise as details to be kept. 
Nice topic Btw 
Carlito,
Presently, TVAI doesn’t have much in the way of color control. - (Hopefully one day it may have that and more.)
My basic approach would be to use a video editing application that has good color controls to modify the video source, or the output from TVAI.
If you aren’t talking about the noise that’s generated when a video camera is use in very low-light conditions, this also might apply, depending on the editor you are using.
I sometimes, use TVAI to denoise the video at its original size, saving it into a high bitrate format and then using the video editor to modify the color. - Sometimes a little desaturation will give TVAI more detail to work with when you take the editor’s color-adjusted back into TVAI for enhancement and resizing.
1 Like
Meanwhile I’ve tried Gaia HQ and problem solved
I just don’t like some enhancements done by Gaia but since I’m working with Tif files i only need to replace the source frames where the chroma noise appear with the cleaned frames generated by Gaia.
Enjoy the weekend and Thanks 4 the help.
It would be a nice feature is they were to drop a low-light digital noise filter into TVAI.
So much of the most recent video is being shot using digital cameras and unless you have one that is set up for use in low light, the noise will most certainly be there.
Notably, most of the modern of the still photo editors already have this kind of denoising built in, it should be an explicit setting for pre-enhancement use in TVAL
Over the past 3 years I’ve used VEAI I’ve come to realize that Proteus is not a good general purpose upscaler. results drom SD to HD are never great no matter how much tweaking of the knobs you do. It has a place with >720P - 1080P video as it can clean things up and not wash out the details if the source is already high enough. I mostly use it to clean up low bitrate 1080P video outputting 1080P
Gaia is far better for clean SD content but if the SD stuff has compression artifacts they turn into noise with Gaia. So the other use I found for Proteus is to just use some decompression and denoise with nothing else at all and keep the same resolution. Basically just clean it up a little before using Gaia. Too much denoise or decompression and Gaia has nothing to work with and it basically doesn’t do anything.
A good addition to this post would be for someone to post a sample video and whoever is interested can post a result trying to fix it.
hello, for the moment I am there, using hybrid first (gtgmc, denoise dehalo sharpen and line) then with veai (themis + appolo soft + artemis medium in a single pass (production in 720x 576) to finish with vegas pro in hd (1280 x 720) if anyone has any ideas to point me in a better way… 
because when i activate themis, depending on how i set the hybrid filters, it gives me a series of distorted images so much the deblur is pronounced at one pass: - YouTube
TVAI and all of the enhancement methods are parts of a tool for enhancing video. Depending on the methodology you use with any of the enhancement methods, you will get good or bad results.
Personally, I get very good results using Proteus, especially if I use the ‘Relative’ option and I am careful in making my settings. With this, I can take a lot of old very low resolution video images, clean them up, and enhance/rescale. In most cases, it would have been impossible to get the quality results without TVAI/Proteus. (A few years ago, there was just no way, period)
If you aren’t getting satisfying results using Proteus, you may want to look at your methodology. -
Beautiful sculpture can be made using a hammer and chisel, but those same tools can be used for demolition.
1 Like
Nothing new here…
I think this software is now just being rewrapped with a new interface, having it support parallel processes automatically now (instead of us having to do it manually) and it is just starting to use resources provided to it properly and efficiently now. So of course it will run faster.
These other “new” features just muddy up the software. They are adding things that I can get elsewhere that are proven bug free and use resources efficiently.
Topaz Labs should pick something to put all of their focus on and make it better than anyone else. Instead they are trying to be an all-in-one application. In my world, that will never exist. I buy plugins from many different companies that perform the task I need and that do it the best.
I never thought I would be buying software that I would have to wonder if the next version would work or not. I have never seen this in the industry. They are cheapening their brand and it is very unethical in how the past year has been handled.
Yes - they have added some now models (I guess) – but I have not seen a dramatic increase in quality since 2.6.4. No game-changers here.
All I see are little tweaks, little sharpens, extra noise, less noise, extra grain, less grain – but overall – the images I am seeing are not dramatically better than 2.6.4.
I think this software has hit a wall on actually providing new “wow” developments in upscaling technology.
I keep seeing people compare images from minor versions that are just slightly different from each other. It’s a never ending battle here. I feel like I am getting an eye exam – whats better image A or image B? A or B? Ok how about C? lol I mean seriously – a 5% change in quality is not a game-changer and everything used for comparison is not standardized.
It is great they are fixing bugs. That is expected.
I now feel cheated by their 1 year of so called upgrades. There have been no upgrades, just fixes and extra fluff that I did not purchase with 2.6.4. So v 2.6.4 reached its EOL last year. I mentioned this to Topaz Labs and they remained silent.
It is ridiculous that they include this newly designed application as an upgrade, v 3.1.0. It has a new name and and new look. It should be v 1.1.0. Instead it has all been very deceptive.
When I look at what 2.6.4 does for me compared to 3.1.0 – there’s nothing new that says – OMG – this will revolutionize my workflow. Instead I think – well duh – glad they fixed that.
AI is changing so quickly, Topaz Labs is going to quickly lose their market share, unless they get their act together and decide if they are committing money to R&D or just making due with what they have now.
I feel v3.1.0 is what I should have received when I purchased v2.6.4.
These forums get insane. For people that rely on upscaling software in their industry on a regular basis – it’s ridiculous how all of this is being handled and it is becoming a waste of time and money.
Topaz Labs has lost credibility in making me pay more money to fix the software I have already. I am glad I did not buy 20 seats.
1 Like
Maybe model quality has not improved, but performance does. And that is very important for high end systems.
My “methodology” is fine, it’s just to not use Proteus when upscaling, unless as I said, the video is already 720P or greater. Everyone who seems to love Proteus always gives you the whole “it’s an art, ya gotta know how to use it right” thing. It’s just a bunch of sliders, it’s not an art.
Proteus is good at reverting compression and has a decent denoiser, the sharpeners are pretty much garbage, though the de-halo filter is pretty useful. Anything that you’ve done with only Proteus I can pretty much guarantee I can do better with (most likely) Gaia. The trick is to run a non-upscaled pass with Proteus just to deblock/denoise
I have a hard time telling the difference between Gaia and a Lanzcos filter upscale.
The new versions of most Enhancement methods have been tweaked, to different extents. Personally, I have found Proteus in both the standard and new betas behave a little differently and as, such, the approach to adjusting them must change, too. And I made some changes to my approach to settings and got results that were as good or better than before.
If the controls for an enhancement act differently, you need to become familiar with the new setting it requires. - It’s like buying a new car. The handling is going to be different…
1 Like
Proteus does wonderful upscaling. I use it almost exclusive ly for that. However, the input must be always cleaned up at original, but not so much as the ‘detail’ gets cleaned out of it, too. (Proteus can do this too, If you’re careful to not overdo.)
FYI: after a careful cleanup at original size and using a low loss/lossless output format use proteus-Relative to upscale, in increments. If your source is 720, go to 1200, adjusting carefully. That gives Proteus AI to use the original image to build out the ‘clues’ it found in the original image.
IMPORTANT: This small upsize doesn’t force the AI past its normal limit. If you crank the decompress, denoise, sharpen up settings too high, it will ‘accommodate’ you by ‘faking it’ and it makes a mess!
Repeat the Proteus enhancement to full size or another intermediate step. Once again, don’t push the settings.
The final step: Upscaling to full scale, you can finally crank up the sharpness and detail to give you a very clean, detailed result.
Sure! The multiple passes do take longer, but it makes a difference that’s really significant.
3 Likes
I have not seen a performance enhancement compared to running multiple instances.
They have just done what should have been done before automatically.
Nothing has changed in rendering efficiency - they are just now properly using the resources correctly.
1 Like
I disagree 100%. That has not been my experience using the same video clip from 2.6.4 until now. I have images that prove this to be true.
If you post - back it up with facts.
Precisely! Which (shameless self-plug) is why I made this topic:
Some pre-processing is simply best left not to A.i.
2 Likes
I agree 100%. Give TVAI your best video possible. Then let it do its thing.
There are many applications out there for PC and Mac to accomplish it.
TVAI should have a minimum criteria posted on what it will accept — for best results.
1 Like
Although I do have a few outside utilities I use for massaging video into good shape for TVAI, I have noticed that several of their tools for deinterlacing and clean-up have improved significantly. They have become much more useful as a result.
In any case, never denoise or decompress to the point where detail degrades or the image gets distorted. And always at original resolution.
2 Likes
That’s all going to change fairly soon; although that also depends on the hardware you’re using.
Interesting factoid: The cost of 4TB high speed NVMe is dropping. Also, there are lot of new adapters that are evolving to provide capability for having up to 1000 TB of NVMe SSD on one machine. - I’m sure it will be few more years before they get the wrinkles out of that last ‘fact,’ but it’s here.
