Sorry for the question, but I am confused by this. Before you had your original video shown. Then you would apply a setting. After that you can always compare the original and the newly rendered version by split screen, etc.
I do not understand #1 and #2. How can I just have it show one window that I can scroll back and forth between the original and the rendered version?
It seems so complex now. Perhaps I am just getting old. I find myself taking more time to setup videos to render than before.
The point of 1 and 2 is to apply different settings and compare them between the two so you can select the best settings. When you export there is no longer a live preview.
Topazâ AI fully works only on your computer, so no privacy issues
And how is that a selling point for people with older or less powerful devices that donât want to spend thousands of dollars getting new equipment, in the hopes that one program âmightâ work?
Itâs not - and I never said that. Just that itâs a different approach with advantages and disadvantages.
Not every one feels comfortable (or is even allowed to) load up personal images / videos on a third companys server.
Besides the local approach is more flexible - options like output as a series of PNGs or low to uncompressed 4k video isnât really feasible with an online approach even considering todays internet speeds.
I tested TVAI and another video AI program on a cloud-based virtual desktop a few months ago. The reason was that my main machine is 100% Linux and I prefer to do as much as I can from that device.
The cloud desktop worked fine, only problem was that I wanted to change some system settings and the company did not allow users to adjust the settings. And, like you said, they slipped some language into the ToS about having the right to look at what youâre doing.
But the future of all mainstream app development seems to be to include at least some server connectivity. Heck, even TVAI checks your license through a licensing server each time you launch the program instead of using offline licenses. And you have to manually download the models if you want to keep them on your device.
And thereâs also the fact that Topaz collects anonymous usage statistics by default and you have to manually turn this off.
Iâm in total agreement with you. With Adobe, there is no privacy period with any of their AI products. With Topaz, you donât really have to worry about it since all processing is performed locally.
I agree with @michaelpfost, I donât believe Topaz would be able to compete under their current trajectory. âBetter privacyâ is not a strong argument against âbetter stability, performance, and UI/UXâ, especially now that more established companies are getting into AI.
But Topaz seems to understand this, and theyâve already begun integrating into other products, for example with the release of the Ae plugin.
Thatâs really no explanation.
Of course the needed models have to be downloaded, otherwise TVAI would be a 16GB+ download for each version. And those are Topazâ data/code that you are downloading, not the other way round.
Besides, if those models are there, they donât need to be downloaded again - fortunately as otherwise I could not exchange the current extremely slow Iris model files with faster ones I extracted from past installations (a BIG thanks to TimeMachine here).
So the internet connection is really only needed for activation - the actual encode is fully local: Iâve done quite a few encodes with capped internet earlier as I feared TVAI would replace my faster old models with new ones and thus took the rig from the net - until I noticed that it doesnât do that if it finds an appropriate model.
And last, not least: Iâm not talking about âsome server connectivityâ or anonymous usage data but about uploading your private images and/or videos in total to a foreign server that often is located in countries with dubious user rights.
Also, it doesnât even matter if you would accept that as in quite some cases with professional work you just arenât allowed to upload images of your customers (at least not without getting explicit permission by them to do so). Maybe in the US this doesnât matter but with the european GDR you can have a big problem here.
Of course, the online approach of e.g. Adobe has itâs advantages for both the user (that doesnât need a strong computer and still gets faster results) as well as the developper (who has an exactly defined setup and thus doesnât have to take care of those many many different configurations with different driver releases, making software development MUCH easierâŠ)
The same could be said about Google Drive, One Drive, and so on. It doesnât stop anyone from uploading copyrighted material to their servers. Even YouTube lets users store copyrighted content on their servers, it just restricts viewership based on location.
