Example:
Panasonic has an official H265 license. The H265 video files of the GH5 caused many problems with various NLEs. The licensor didn’t care, Panasonic paid for the license.
Panasonic also has an official ProRes license. Apple has strictly monitored and checked that the ProRes files that came from the GH5 were 100% compatible with other ProRes software. If that had not been the case, Panasonic would not have been allowed to sell its cameras with ‘can record in ProRes’.
And no, if Apple were to turn this into a lawsuit, no court would say that Panasonic is free to sell its own, not always compatible, version of ProRes as ProRes, even if they have a license.
Thank God! That a camera can - if the customer requests it and that happens quite often - record in a high-quality, very precise known format that almost everyone later in production can read and further edit without any problems is a primary requirement for many camera operators. For that, ProrRes is the only serious option.
Until of course a judge bans Apple’s iron grip on ProRes and every manufacturer may user its own home made usually-compatible-but-that-requires-occasional-some-testing-and-using-the-correct-version variant of ProRes…
Constant random crashes under “Pro” version
It can batch 5 files flawlessly and blast everything following
Either multiGPU on one file or multiple files on multiples GPU
The load behaviour is confirmed completely erratic.
My “Pro” license is USELESS, that’s all I will say.
I will not waste my time with logs or investigations, I did it enough for the bucks spent. There are enough CUDA errors that can be patched simply “using” the app., thing that dev. do not seem to do before releasing.
Wow! This report really saddens me (especially after the so-called promising Tensor uplift Studio driver 561.xx). I was just hoping… Oh, well. It is what it is.
Thank you, Alpheratz, for your continuous rigorous findings for these multi-GPUs results.
This highly does not come from drivers as I tried many of even the top latest ones.
CUDA errors from the log that are pushed like a bull to GUI as “Error” may need a accurate analysis.
As a reference point I always go back to 5.0.4 to figure out if any other parameter ruins things, and oh : miracle … 5.0.4 runs like a charm
This behavior along with the rushed „renew cheap now now“ continues to manifest the impression that topaz has a great start and instead of improving importing innovation and models, has shifted to a squeeze them business approach.
This points me towards heavily considering getting a plan b ready, as we most likely will not move towards the new Eula.
I just got an email that says that if I move to autopay I can get a discount. I like the product and I will sign up for autopay AS LONG AS THE MODELS ARE IMPROVED MONTH AFTER MONTH. At LEAST every 2 months.
I won’t subscribe to receive bug fixes and layout changes. Do you understand me? This is not a charity.
They have not fixed the EULA yet either. The current luangauge makes it a subscription service and the luanguage someone linked, that was being considered as a fix, still includes the rug-pull clause in the amendment section.
Sorry to say, but this kind of business method is a bit like the Tesla model.
And Tesla is getting a serious backlash, after years of questionable hype.
You pay for something you may or may not have, and it turns into “Hey idiot, you’ll never have it! I got you”
Either way, and I don’t wish this on Topaz at all, but if they go deeply down this wrong path, the market will kill them.
The reason they need the money so badly is mind-bogglingly obvious and… worrying.
I don’t know anything about the company, but they will have to deal with some judiciously and well-enumerated clarifications, in my opinion.
What is absolutely incredible is the total lack of communication from them on the forum. There is an armada of disgruntled people and responses that can be counted on the fingers of one hand.
The whining about models has reached a new level! Now, Topaz has to release improved models every 2 months… or else!! I think to achieve this, the training bots shouldn’t be allowed any time off. Yes, they’d like to see their loved ones just like everyone else, but this lack of new and improved models is the final straw! Forced overtime, no meal breaks, and start cracking the whip!! If they don’t get with the program, there’s plenty of other bots to take their place! They’ll be the exact same bot but I don’t care!! More models, now!
Live render not working with iphone portrait video… (the live render comes out unrotated and stretched), tested with Rhea but I think it happens with any enhancement.
IMO it’s inevitable that every model eventually reaches a point where no further improvements are possible without changes that are so radical that it becomes a new model. So we get something like Rhea, which Topaz has described as a cross between Proteus and Iris, while updates to the earlier models slows or stops altogether.
I find that when I set the codec to ProRes 422LT and export multiple videos, only the first is exported with this codec. The 2nd video onwards all return back to the default settings in the preference (which was h264 high).
.
