Updated preferences to be able to disable tooltips
Fixed grain only showing for progressive models
Fixed intermittent login issues
New in 5.3.0.1.b:
Fixed:
Mac window controls now display correctly
Timeline playhead no longer freezes intermittently
Timeline now updates during playhead drag
Changed:
Added additional margin to export progress indicator
Fixed āSave & Restartā not re-opening project on relaunch
Added video input information (FPS, resolution, pixel aspect ratio) to a hover panel in Sources
Fixed extra frame output in Enhancement models
Fixed auto-parameters not loading intermittently
Control panel updates
This update includes a new Model Discovery system in the right panel, which shows examples of when to use each Enhancement model:
In addition to the new carousel, users can also switch to a list view for more compact controls:
This new model selection tool will help users navigate our growing collection of AI models and find the right fit for any project. Weāve also added a star icon for models that have very broad applications.
Timeline updates
Video AIās timeline now offers much more functionality, with frame-accurate timeline markers and a new system for timeline zoom:
Itās really nice, if it works fine but this is a bit for ānoobā. If you read the tooltip you have all you need, doesān it ?
Somehow what Photo or Gigapixel does, it can inded save a lot of time with searching !
Timeline upgrade is cool !
Donāt forget sir to follow your roadmap
My first feedback was : in A/B when you select 2 presets and you donāt touch them why the preset combo box is reset as soon as you switch into A or B ?
For advanced users, we have kept a compact list of Enhacement models. Video AI is growing each day, so some features will be more oriented towards newcomers.
Weāre also very excited about the timeline revamp, definitely feels more functional when working with long videos or trimming very precisely.
Good point re: presets. That text box should show the preset name until a setting is changed ā we will look into this for the main release next week.
The argument is that advanced users wonāt be harmed since the list mode remains available, which is fair. However, I donāt believe the new discovery mode will benefit anyone, as the short video clips presented donāt bring any information even though you can see comparison between original vs result by mouse hover.
For instance, if youāre trying to decide between the Iris model and the Rhea model, the sample videos for both are human faces. What insight are you supposed to gain from that? Should the presence of leaves around a face indicate that you should choose Rhea? (This is a joke, but you get my point.)
What value does the flamingo video on the Proteus model adds? Itās hard to find any differences among those AI models in those short video clips.
The most important thing is not āusers can also switch to a list viewā, but displaying something informative and relevant for someone.
Showing just āniceā short movie has nothing to do with discovering the appropriate AI model.
Do you mean like a short blurb about what the intended effect would have and maybe a little short list of typical use cases? Iām juxtaposing in my head here that rather than answer a visual inquiry with a visual answer (which some may pick up on and some wonāt) that perhaps the targeted audience though creative is also likely to have some level of analytical techy thought process that might benefit from a textual description appealing to our reasoning and understanding? If so, I think I might agree with your premise here. Both can be ideal I suppose. Just had a thought. When I use adobe, though I often find myself returning to learn about some new or unknown to me functionality, at first glance when launching an updated application (think when update is significant enought to where they reset the pop up tutorials), I find it very annoying. I do like, however, a list of features and recent updates with a description. If I find myself struggling at some point, rather than go through a list of new but unrelated features, Iād rather have an option to click into the text description for other supporting resources which then could be a video example. The cool idea (for me) is if I could set in preferences across all applications a choice of present new information pictorially, textually, or given many platforms use it, there must be an audience that benefits from the autorunning video example. But for me I find it an annoyance. As ubiquitous as it is, I could argue that type of setting could even be a good add on to say Windows itself. Much like Dark and Light can be set and your applications at default follow the system setting. I actually really like that method of setting preferences at the global level and then perhaps depending on the application or my use case having the ability to overide at the ālocalā level. (for lack of a better word). Macro/micro may have been more appropriate.
I easily found a BUG that makes the video timeline scrolling experience worse than in the stable version, I show you in this video:
Basically in the stable version, you can (if you hold down the left mouse button) easily scroll through the video previewing all the āframesā of the video, with a smooth experience⦠while in this new BETA version, if I hold down the mouse button in the same way and scroll through the clip in the timeline, I donāt see the preview of the frames in the player: to see them, I have to click on a point and release⦠and this makes the video visiblity really worse compared to the stable version!
Just tested few short videos with new beta. It looks really good.
I love the new Model Discovery because it is visual and easy.
I noticed that you hide Parameters section and we need to enable it.
I love this way because I am using most of the time the Auto parameters function.
So for me, it is a great improvment.
About the Timeline new implementation, I love it too!
Also the top tabs Adjustments / Codec settings is at the right place I think so! Great that you take the time to organize your interface like CAD software with tabs at the top!
The only issue I noticed is about playback. When advancing frame by frame, sometimes, I have no update or takes a while. The playback is globally less responsive than current official version.
About rendering speed, it feels great. With my 10 years old NVIDIA 980M graphic card, I upscale from SD to FHD at 2 FPS ! From HD/FHD to FHD at 0,7 FPS.
I totally agree and I point out that I am seriously and with no humor at all when I say that the new model discovery system only enlarges the application unnecessary.
Personally me on my side I find that focuses too much on the interface of the software than anything else. Developers should focus more on model updates be it Old or new because there itās been several months we have more updates on the models. I find that the new RHEA model and still as slow to export video it would be good if it is a little faster I donāt think Iām the only one to have asked for this kind of thing. For developers why you devote much more to the software interface when we As a beta tester we donāt care about the progress of the interface, what interests us is the evolution of the software models, their update and that these models are whatever From above quality to export video and solve the various problems that are cited by other beta testers on the forum. The priority is on that but not on the interface.This is just my opinion.
I am talking seriously without any humor to you when I say that after loading one of my self made presets it did set the wrong model and forgot all enhancement settings.
As for playback speed/responsiveness ā we are making fast progress on an entirely new video player in the app and we think youāll be very pleased with the increase in speed when navigating through videos, especially 4K+ or highly-compressed inputs.
The timeline revision is good. Iām not in the habit of scrolling, so didnāt notice the issue flagged above.
Less enthused about the carousel. The visuals are too small to be of any use in displaying the relative benefits of different models, so they just end up as background art behind the text descriptions. IMO it falls far short of the old four panel preview that compared enhancement of a few frames of the actual source for different models.
Can we please add ājump forward/back to next minor and major dividerā buttons so I can accurately jump to each line? Iām referring to these lines:
Also we need the ability to add and delete timeline markers and to jump to them as well!
All due respect to the devs/researchers, I donāt know their schedules, but yeah, I have to echo this sentiment, as Iām looking forward to new models featured in the betas not so much for the UI; however, this is not a demand on my part, just that I see a lot of new models from TPAI make it over to TVAI,
As I canāt wait for that tech to make it over to TVAI, imagine, a way to erase backgrounds/people/things from the video would be incredible!