Topaz Video AI 6.0

Topaz also has a perpetual license. I hate that word because it’s misleading and/or misunderstood. But Acrovid uses the same business model as Topaz in this case. There are a lot of competitors out there but only a tiny handful that truly compete. I do believe that I currently own all of the (paid/subscription) ones that truly compete which includes: Topaz Video AI, Neofootage AI, Unifab, VideoProc Converter AI, and then a couple of NLE’s such as Davinci Resolve and Vegas Pro. Video editing/upscaling is more than a hobby, it’s a lifestyle :grin:.

Everyone I own works great for it’s intended purpose, but some are better than others for sure. Some are obnoxiously priced while others are very to extremely fair in cost. But the single program with the most potential in my opinion for specifically repairing and upscalling old movies (that’s what I do), well it’s Neofootage AI for the win!

Topaz has gone to hell and looks like it will take an act of GOD to bring them back. So keep with the devil you know or play the field a bit, that’s up to all of you. But I do intend to make a very comprehensive comparison video between everything I currently own on my YT channel. Look me up as ( Now Review This! ) if any of you have any interest in such a thing.

9 Likes

What makes Neofootage AI better? Might be looking at options since the dev team here seem more focused on GUI (which is going the wrong direction as well) than what people buy the software for.

4 Likes

Well, in 2 words, pure capability! It’s free to try, there are plenty of videos available to know what you’re getting into. I don’t know what more to tell you other than check it out, or not.

3 Likes

Right on, thanks for the clarification :slight_smile:

Seem like topaz labs keep shooting themselves in the foot. So glad I didn’t renew my license. They can’t just add new models and fix bugs instead of introducing new features to sell every year. Consistency is key here, stop overhauling everything

11 Likes

I’ve said it many times, but you guys need to stop doing UI design for a while.
Seriously, everything is a mess.
I don’t know who’s in charge of UI design, but seriously, most of your designs are so uncomfortable and have crappy usability.
Please, please, I’m not trying to stop you guys from being businessmen, but at least balance functionality with design.
I’ve been a subscriber to your product for 2 years, but I think I’m going to have to stop now and go back to pre-v6.

11 Likes

How many new bugs were added in this major update?
It is very disappointing, even though your technology is excellent.

1 Like

What is better about tabs? Give me one example please of how/when/where such a thing is useful in a program like this. I seriously cannot think of any situations where I thought to myself: Gee, if only I had a new tab to clutter my workspace, life could be so much better then?!

5 Likes

VFR bug still not solved???
Dude…

6 Likes

You deleted the old post and posted it again—yet you kept the false statements in it. Why?
Processing speed has not changed. (Some people report it’s slower, but Topaz has not stated that they made any improvements in speed for any models.)
Image quality is the same. None of the models have new versions. No new models were introduced.
If you get good results in TVAI 6, you can get the same results in TVAI 5.

11 Likes

Pictures, or it didn’t happen.

Indeed.

5 Likes

Oh, I checked it out. Even bought a license, LOL.

It has decidedly less models than TVAI. That’s undisputably true. As for ‘capability’, that’s a tough one. because capability is closely related to how effective one can use those capabilities.

And their UI is, not a mess per se, but like entering one of them old nuclear plants, and looking at a wall full of buttons and lights. :grin: I was a bit turned off realizing it will be quite a steep learning curve. Then again, DaVinci Resolve isn’t exactly easy either. Still, (pre-6) UI of Topaz is decidely more intuitive. Neofootage AI has pretty nifty color grading options, though.

I told Topaz they shouldn’t make victory lapses yet, as they mere fact that people are checking out the competition, should have them worried: even very loyal customers, like me, will eventually leave them if this keeps up. Caveat venditor, as it were.

5 Likes

Poisons, ALL of them! But you have to pick at least one to do the job. Nobody ever said you have to pick ‘only’ one though.

Comparing Acrovid to Topaz is a bit of an apples to orange comparison. Topaz is trying to be a user friendly everyone can use it with minimal effort kind of a thing. Whereas Neofootage AI is specifically professional with the learning curves one would expect, but it has real professional capabilities. Neofootage doesn’t seem to hallucinate like Video AI can/does. It’s more of a non-destructive program out of the gate, but to truly benefit from it you have to learn the program and it’s settings/capabilities.

When it comes to mapping SDR to HDR it’s just no contest, Neofootage and Davinci are the obtainable options with Colorfront Transkoder being the king and demanding the price for it’s place on the mountain.

But upscalling is nothing all that special, going from 1080 to 4k is easily achieved for free. What Topaz has and the only thing they have currently worth value is their AI models. As much as Topaz keeps destroying the user interaction with those models, they still have the best (all in one) variant that I am aware, but not by much and certainly not for long. I have it on good authority the next generation of models are coming soon to Neofootage AI which will further close the gap, but not the learning curve.

You simply cannot shortcut an education, nor should you underestimate the cost of a good one. Wanna take a wild guess what I have invested thus far just in $$$ not to mention time?

3 Likes

ISSUE: Before and after don’t match frame to frame, there is a shift in timeline
Please fix this, thanks!

4 Likes

ISSUE: there is a color shift between before and after. It is easier to see with orange, red, pink colors in the frame

magenta-red hues are shifted towards orange

8 Likes

It’s your opinion and thats ok, but if the V6 Output would result in a different output, even when the models doesnt changed, and it was not announced, then that’s nothing that should be considered positive :smiley:

Why ?

Lets say you have a freak like me^^ that does restauration from officialy released SD pron^^ DVD’s and want to >share the way how you did it< (or want to do it, in a few years again, because Post Processing Tools get better) then it’s bad if Version X behaves different as Version Y, and nobody informed you.

The List in 5.X was not perfect, but at least, i was able to loop. Compare the same frames (if you push Cursor Right, active Frame from different Entry in the List, will be the same, after you click the other Entry. Same with a Duplicated Tab in V6, which is Good ! thumbsup ) from different previews, just by clicking on a different Entry.

Now before i tried V6, i was thinking “hmm, with tabs the TreeList in 5.X (that is not optimal) is not needed, if they add a pure Frame X to Y Joblist (with Join ability), and i can compare the output Frames from different tabs by just clicking a Tab”

But the reality was, that there was no Loop, and no Info, which Frame Range (Exact Start Number and End Number) which Range Processed, all what i saw was the active Frame.

Ok, so i duplicated Tab A (still not knowing which Frame Range) then tried different Settings.

Moved by Cursor Event one Frame forward, Clicked TAB B, same Frame ! Good !

Now ! TAB B was active, i clicked on Split View. TAB A was now left (and directly right from “Original” without any Space between) and TAB B right.

Ok, then i switched back to Single View (do >NOT< Click X on the right Tab ! it will be closed, without Question) Clicked TAB A, clicked on Split View, and the Result was, that the Original Tab was now Left, and TAB A & B on the Right.

That’s inconsistence behaviour.

After that i clicked Single View, and guess what ? It doesnt worked, i had to Click Fullscreen (that fancy stuttery slow slide in and out animations for the bars look amateurish and consume time ! Fast Snap in and out would be 100x faster and better) first for some reason, then again, then Single View, then Single View was executed.

So, how do i know which Frames were executed from Preview ?
RClick on the Preview, then RClick Menu will give you the Info, but just if you are on a Frame that is allready processed. (So you have to Click on green Spot on the Seek Bar first)

Ok, lets say i want to throw that Range of Preview Frames to the Export Queue (and dont want to start. Because it’s allready Ok and Processed !) then i can forget that.

I have to RClick first on the Preview Window, get the Frame Range.
Seek to Start Frame, Mark In, Seek to Stop Frame, Mark Out, then start the Processing
(so i got no other Choice ! No Option ! No Output Joblist, where i can add “Frame X to Y” Jobs inside, modify the order from entrys like i want, and start the export when i want) by Clicking “Export As” and later i got a Ton of Files with stupid names like.

Misty (TopFieldFirst)_1_ddv3.mp4
Misty (TopFieldFirst)_2_ddv3.mp4
Misty (TopFieldFirst)_3_ddv3.mp4

Nothing was joined ! Any Entry in the Queue List, was processed as a unique Output File (FFMPEG,VirtualDub1&2,AviSynth can join stuff without problems)

There is not even a Option for that !

Why should there be a Option ? Flexibility ! For situations like that.

Frame 0 to 1000 is good with actual Setting A
Frame 1001 to 1003 messed up skin on the legs, need other setting. Call it Setting B
Frame 1004 to 5000 good with Setting A
etc
etc

I dont want to ask my self, which Output File is for which Frame Range, and to join them externaly, when it would be easy to integrate a real flexible Joblist with a Join Function.

I dont want to Process a File 5 Hours, just to see afterwards, that Frames 1001 to 1003 are messed up, so that i have to Split the whole File to File A Frame 0 to 1000, File B 1000 to 1003 etc etc etc !

Amen !

2 Likes

CRITICAL ISSUE: When exporting a portion of a timeline, the output video starts with a black video and original audio.
The video content starts later on. There is a misalignment of the video and audio in the output and it is about 8 seconds for the 17 second test fragment of the timeline.

Edit: this is occurring when Mark IN is not at the beginning of the video.

1 Like

i have this issue aswell…

2 Likes

Maybe, maybe you are very lucky and you are the only person who has a computer where Topaz Video AI 6 works properly.

2 Likes

There’s a lot being lost the the language translation, for sure. My best guess is they think that saying something nice about it will get Topaz to work on their requests.

1 Like