Bot or Cluster B ^^ guest
One post (that he deleted) statet that the text was created with Chat GPT.
Here in germany most normal people associate it with bad xD
1 Like
Vielleicht nutzt er chatgpt zum übersetzen und schreiben.
Was is bei uns so falsch an grün? 
Habe gelesen das es möglich ist KI modelle durch unsinnige fragen dazu zu bringen sich zu enttarnen.
Das machen z.b. leute auf reddit.
Du fragst es dann z.b. ob es dir ein gedicht schreibt.
Maybe he uses chatgpt to translate and write.
What’s so wrong with green (in Germany)?
I read that it is possible to get AI models to unmask themselves by asking nonsensical questions.
People do that on reddit.
You then ask it to write you a poem, for example.
4 Likes
Eh! mein englisch denkendes hirn hat das nicht verstanden.
Aber ja. 
2 Likes
Another “revamp” ? Every time I get used to a user interface, your developers throw it out for a new one which apparently everyone’ is asking for? Well not me. As for “We finally listened to you!” ??? What? I’ve read back over the forums and people repeatedly are asking for stable interfaces to remain stable. I’m on version 5.5.1 and overall, it’s working for me don’t want to start all over again with a new software interface. Enough! Seriously or you’ll lose my subscription.
There’s nothing wrong with evolving this program, but please sandbox them for a while in stead of using us paying subscribers to test your programs for you. Also when you are rolling a new interface out, allow us to run the old version concurrently. At the very least, we should get extensions to our subscriptions to make up for all the lost time trying to get your new versions to work.
Also You really need to provide support for older stable versions ongoing
7 Likes
I know this as:
There’s 10 kinds of people: those that understand binary and those that don’t…
There’s 10 kinds of people: those that understand Qubits and those that don’t…
1 Like
Yes. For instance, ask it to count the number of times the letter “R” appears in the word “Strawberry”.
That’s because ChatGPT is a scam: it has no intelligence, nor does it really understand your question. Instead it tokenizes your sentences, and thus cannot process simple questions like these.
2 Likes
Btw, this works with any word with repeating letters in them. E.g. ChatGPT will tell you that there are 2 “r” in the word “error” and insist on that. Gemini says there’s only one “r” in “error”, so it’s maybe even more dumb.
4 Likes
I had to test it out
2 Likes
When it comes to error, the only thing that comes in my mind is this
4 Likes
troll: “All your error are belong to us.” 
3 Likes
I see in other responses that 5.3.6 was the last time Video AI didn’t change your video to VFR. I have been using many hours pulling hair out over several weeks trying to figure out where in the chain I was going wrong. I am now installing 5.3.6.
Further (and independent from the point above): I had just gotten used to the GUI in version 5 and really liked it, but now it’s gone. Such drastic changes in GUI may be happening too soon.
4 Likes
I’ve been watching and reading this forum since version 2, and still have not seen sufficient evidence that this software is much different than what it was several years ago. It seems to be the same ole complaints year-after-year which makes it impossible for me to spend a single penny for an upgrade.
Speaking with my sister, who has worked for Adobe for over two decades, told me in a conversation about a few engineers she works with who believe that AI upscaling is still in the gimmicky phase, with many companies selling garbage software that amounts to no more than shareware from decades ago. She laughed when she said Adobe is many years away from highly refined algorithms which can upscale low quality video with stunning results using tools like Premier. The technology today cannot deliver quality results, and Adobe refuses give themselves a bad name by joining other companies who think they have the magic wand that can transform any video into something that looks like it was originally filmed in glorious 4K. When I asked about VideoGigaGAN, she was told that it was only proof of concept and many years away from reality.
4 Likes
This is a purely competitive troll opinion.
What is the threshold where we should say “wow, this is revolutionary!”
Obviously it won’t happen overnight but gradually.
What makes me laugh is that we’ve been hearing the same speech for 20 years with video games.
Take a Gran Turismo 7 on PS5 Pro and I challenge you to tell the difference between the rendered scenes and the filmed scenes
And all this happened gradually, without ever going back
Nevertheless, despite the pile of unbearable bugs everywhere, TVAI is the best there is, at least for the general public.
So Adobe just has to do better instead of laughing with its gigantic means.
Many of us are still going from 1080p to 4K and seeing it as a real gain, despite what Adobe claims.
- Rhea is 90% awesome, but requires crazy tweaking to get the resolution you want
- Rhea XL is exceptional “sometimes”, some textures are recovered well, while others are ruined
- Hyperion works well and is welcome, it’s 70% awesome for me
- Iris is a good second pass to Rhea
Everything else is kind of a throwaway (in my opinion), but that’s already huge! What Adobe just doesn’t have
9 Likes
Dont underestimate Proteus V3 and Artemis.
Proteus V3 can do sometimes marvelous things (especialy when it comes to Hair)
The only real drawback is, that it works Single Frame based (Spatial)
that it is not good at retaining the Original Form from Shadows (important for stuff like subtle Fatpads on legs, *utt etc)
If Artemis (works Temporal & Spatial) works good with the Source it is very good when it comes to deblocking, denoising, debanding, resharpening & retaining the Original Form from Shadows.
In 2.6.4 when it was used with Grain and the right Grain Size. Then Artemis would keep/restore more Details that match the Size from Grain ! (maybe not by intention from Topaz) that’s another reason why i want the Float Numbers for Grain back.
Proteus Manual (pure Spatial) and Auto/Dynamic (pseudo Temporal) are 2 differents things.
They use both different Models. I saw that yesterday. After i used Dynamic, i switched to Manual (no other change) and Video AI startet to download a Model.
2 Likes
I compared versions 5.3.6 and 6.0.0. I did not see any difference in the work for my purposes. Tested Proteus, Rhea, Apollo and Aion. Version 6.0.0 is a few percent slower, which lowers the temperature on the processor and graphics card. I like this more than making a video a few minutes faster due to the increased temperature…
But I’m interested in something else! I didn’t use pause in video processing before, because I rendered small video clips for 3-5 minutes. Now I want to process larger files.
I was surprised to find out that the pause function does not work if you save a file with a constant bitrate (180), H264 or FFV1 in an MKV container. Only if you save in MP4. MOV did not check. And this applies to both versions, 5.3.6 and 6.0.0. Why is that?
Please tell me who thoroughly tested this feature. What are your conclusions? What can you tell me. I would greatly appreciate more information.
I tested VideoProc version 7.9, I saw that it is faster than TVAI and does not give too many effects. It processed 3 times faster than TVAI. It does the effects without exaggeration, I actually liked the new version a lot.
3 Likes
Your sister isn’t wrong. All of us who’ve ever seen small faces get mangled by TVAI, knows there’s no real AI involved. It’s just statistically trained models, that, on a good day yield decent results. Would make for a cute line in Portal: “The AI is a lie!” Thing just is, everyone is fibbing about Ai. It’s just the buzz word of the day; it’s not just Topaz.
I understand Adobe’s position; and there’s something to be said for waiting for true AGI, that will simply never get small faces wrong, ever. As a consumer, though, I’m glad we have at least something, though.
I know, this is not an airport, so I don’t have to announce my departure; but I just want to let ppl know I canceled my subscription today; grudgingly, as I was trying to come up with every excuse in the book to try and stay. With people making/improving models allegedly having left the company, and no working version any more since v5.4, I really see no reason to extend my loyalty any more – especially since reciprocation from Topaz, for months now, appears sadly absent. Who knows, maybe next year. For now, at least, I have a solidly working v5.3.6 I can work with.
6 Likes
What I have now done is taken a lousy 854x480 video that was badly deinterlaced. All models failed when scaling this up, Its low contrasty low details. So I took Rhea XL, skin tones got blotches and the usual bad things Rhea XL does BUT the picture was sharp and the ammount of detail was unbeatable!
Then I upscaled the same bad source with Iris LQ. I had to set Deblur to 100 to make the breaked in pieces pixel blocks (Iris does a lot) bigger more contrasty again. Result was a smooth but washed out few details but homogeneous result.
Then I imported both encodes, which could not have been more different, into my video editing program, overlayed both and set the Iris video to around 65% opacity, i.e. 35% Rhea XL shines through
Export it and in TVAI “proteus V3 manual mode”, Denoise -2, sharpen +8 and nothing else, second model “Artemis Medium Halo V2”
The result is impressive, I would not have thought it possible with this. So Rhea XL is generally not applicable, but “mixed in” it can do wonders. I’ll keep an eye on this and try it out more often now.
4 Likes
The prompts are not correct guys copilot is much smarter
I want all characters with r in one word and all other characters in the other word and tell me which position in the word strawberry you have moved the r s in the word with charater r also make a calculation of each new word and compare this with the number of characters in strawberry as the number of characters both new words need to count as strawberry
Sure! Let’s break down the word “Strawberry” into two new words: one containing all the “R” characters and the other containing all the remaining characters. I’ll also provide the positions of the “R” characters and compare the lengths of the new words with the original word.
Step 1: Separate the Characters
- Characters with “R”:
rrr - Other Characters:
stawbey
Step 2: Positions of “R” Characters
- The first “R” is at position 3.
- The second “R” is at position 8.
- The third “R” is at position 9.
so how many r we have in strawberry
The word “Strawberry” contains three “R” characters. Here they are separated from the rest of the word:
- R: r r r
- Other Characters: S t a w b e y
Is there anything else you’d like to know or discuss? 
