I just upgraded from 3.5.4 and it is slower by quite a bit. This is even after upgrading my video card from an AMD 5700XT to a 6800XT.
I hope this improves with newer builds.
1 Like
It improved for my rig a little (4K upscaled from 1080p): Scrubbing is the most noticable improvement for me though, and that’s not part of the benchmark.
Old:
New:
I don’t mind the AVX2 requirement, I’ll just not renew my update plan, but since it seems to work on non-AVX2 systems with a decent graphics card it would be nice if someone could say if that is correct, or how much will/won’t work if partially true.
Hence people with a lot of update time left and a decent graphics card (or several), as many people here use, would be happier, as if their system would run it fine, except for the installer, then they could use it until that ceases to be the case… (I get it would be a bit complicated to explain to new purchasers…)
BTW I’ll probably use SDF (which I now understand, the docs aren’t great) to do a “proper” 4.0.x install and see how it works with RTX but without AVX2, just not while I’m busy doing stuff with 3.5.4 …
P.S. I’ll also benchmark my AVX2 system, but I suspect it’s so much slower than my non-AVX2 one it’ll make using it for TVAI a non-runner…
Isn’t maybe just changing the gfx card between those systems an option?
Image sequence don’t do anything. I drop a image sequence onto video ai and the cpu usage goes up for a second but nothing. No error or message at all. Its been doing this since 4.03. Has anyone else have this happen? Is there a known cause or fix for this? Thanks.
Yes, please it’s very slow now. I have M2 Mac so noticing real difference with Sonoma.
No slot for a GPU, it’s compact… which is great unless you want to add a GPU… 
Are you using Iris V1? If yes, then maybe I could provide you some older faster model files but that now perfectly do work under Sonoma 14.1 and 14.2.
Unfortunately I don’t have a backup of the fast Iris V2 model files so can’t help with that :-/
Talk to a Mac user here…
1 Like
can you put viewer back like it was so when exporting you can compare in box 2. it seems ridiculous to have to open in 3rd party viewer, thanks
Well, we have limited success.
At least results are not too awful. You can send me a link and I’ll show you results (time-shift+Twixtor in AE).
This is how velocity graph of really bad clip looks:
See the up and down spikes? That’s incorrectly timed frames.
Frames with 0 velocity might look like duplicates, but if we delete them we’d have janky motion, skipping frames.
We create AE script to retime fast-moving and non-moving frames using Twixtor (or Optical flow, which is worse but faster).
3 Likes
Slow motion x2 Apollo doesn’t work together with Gaia HQ or Proteus for me
I will have a look at least it understands the era we are living in generative AI
Can you elaborate, what’s your system and collect logs for support?
Do you select all the images and drop them, or just the first image. When opening image sequences with the browse button, you only open the first image to get the full sequence. It could also be something with how the sequence is named.
It would be nice to see the November 2023 Roadmap since there is a current sale.
1 Like
Personally I don’t like this new version, there were so many things to do with version 3 which was not yet stable, still full of bugs, I will never understand the developers or the management who wanted to release it quickly version 4 which is even more buggy than the previous version. I also find it very strange that they very often release the roadmaps very late in the month concerned. Topaz even stopped gigapixel in version 6.3.3 even though it is incomprehensible this sudden stoppage for several months, in addition they announce nothing, not even an announcement saying that gigapixel is dead or stopped permanently. Gigapixel is much more appreciated than photo ai which is probably not good, and will never exceed the quality of gigapixel it is 100% sure
Apart from that, TVEAI 4 is really not great for the moment, Topaz’s management is really at a low point, I’m not the only one to say that it’s a strong regression for TVEAI. I think that in a few months, there will be fewer and fewer customers for Topaz because it really starts at the lowest end of the scale. I always said that the VEAI v2 team was more suited to making this good software, not everyone listened to me in the past, and finally I see that I was completely right, having changed the team suddenly , it was going to go into regression gradually, and look at the result today, this software was much more stable in version 2, since version 3 this software has never been stable at all, even the last version 3 was still buggy, and now directly v4, and even more buggy.
I would like to say my opinion on this new version, but I don’t see the point, sorry, because everything has already been said by several people, I’m not going to write it, otherwise it would be repetitive of those that we says some people here.
What is the solution today to restore this software to a stable version. Call on the old VEAI 2 team again? change of direction of Topaz? Team change again to a more competent team? To find several people inside Topaz who test the software before making it public? (this will avoid having beta software full of bugs)
It’s not up to me to answer these questions, but I, who have known the software since version 1, will always say, the best team was the one behind version 2. I don’t know about those who are in the forum will say the same thing, but in my opinion, the VEAI 2 team was much more competent than today’s. I don’t offend the developers by saying that, everyone has their own experience at work, I respect them, I’m just saying my opinion on the former team behind VEAI 2.
9 Likes
Isn’t it weird to start working on new stuff when the main application is still in beta basically?
3 Likes
My year is up but waiting for some clearly good changes before renewal. IRIS 2 has been the latest game changer for me, and staying on v.3.5 until
1 Like
“Don’t buy a promise.” Is something I have heard a lot in the gaming community. It usually refers to games that launch in a buggy state or with missing features, but also graphics cards that lunch with poor performance but may possibly perform better after future driver updates.
The general recommendation is “Don’t buy a promise.”
If the game or component isn’t worth the money at the time of purchase, don’t buy it.
I stayed away from 4.x.x until this current version, but I moved to it because it does what I want it to do in its current state. (The previous 4.x.x versions didn’t, so I stayed on 3.x.x)
While I’m temped by the current sale, I’m not going to buy a promise. I will enjoy what I have until the product improves to a point that I want to spend money on new features or performance that exists at the time of purchase.
6 Likes
