Early GUI Design Concept

Hello Everyone,

We are working on a new design concept that brings together elements from Video AI v2, v3 and v4 with a cleaner look and feel.

We’re seeking early feedback to guide the direction of further development. None of these design elements are finalized, and we will continue to iterate based on your feedback. Thanks as always!

Here is a video walk through of the potential new GUI for a future version of Video AI:

Scenes support with different models for different regions would work like this:

5 Likes

Not yet another redesign…

I watched the walkthrough. The GUI is simplified a lot, which might attract the new users but that’s like converting from Final Cut Pro to iMovie.

If this is the direction VEAI is going in, at least add a “Pro” workspace, where existing users who are used to the current design can continue using it.

18 Likes

I agree with you.
But This concept is much worse than “converting from Final Cut Pro to iMovie”, because iMovie maintains basic operational controls without interfering with them.

This concept looks simple, but the productivity is terrible.

The main job is this app is set parameters and convert. Why does it hide the panel to set parameters and require the user to click on it every time?
Overall, the current UI is good, although there is room for improvement in the details.

If the team can only create a concept like this, it is far better to do nothing.

5 Likes

I see a tremendous amount of time and energy spent on the GUI. I truly hope that an equal or greater amount of time is being spent on improving the AI models to get better results. At the end of the day, that’s what really matters. That being said, it appears this new GUI version allows you to manipulate the sliders in preview and see the results of your changes in real time. If that is the case, it could be one of the best improvements to TVAI to date.

10 Likes

I am only an occasional and not professional user and, as with the previous comment, I see that the simplification of the UI is a major step forward especially by being able to see the effect of your changes in real time.

2 Likes

I like the clean aesthetic and the idea of tabs for managing videos, their effects, and exports. Some initial questions and observations:

Is this intended to be the interface for everyone or is it an ‘Easy Mode’? Most of the effects parameters seem to be hidden behind submenus which I fear will just lead to and endless amount of clicking to reveal them. I do hope in the end there is ‘Pro’ mode with all the settings front and center.

Videos Tab

Is this now a file browser or are videos still opened/dragged into the program. Is the left some kind of file tree or watched folders list? The large thumbnails while visually striking don’t offer much information about the actual files. If working with a large number of files a list view would be easier to navigate.

Effects

Will this still include a timeline? The ability in v4 to preview your setting choices, make a tweak, and then click between different iterations has been very helpful for finding the best settings for each video.

Export

I don’t quite understand this tab. Are exports only initiated from the a popup in the Videos tab and the Exports tab only shows the export progress? Having two things named the same in two different places is confusing. It should be ‘Add to Queue’ if it’s not actually exporting. Perhaps look at something like Davinci Resolve’s Deliver tab as a model to emulate. All the settings, initiating the render, and render progress are contained in the one tab.

The export flow needs work.

“Your videos were exported successfully.” This would lead me to believe that the videos had finished but then once the Export tab is opened it’s clear that they are still working.

Thanks for asking for feedback and keeping us all in the loop. I do hope that you won’t shy away from exposing settings and not hiding everything behind submenus. I do understand that from a new user perspective it could be intimidating, but if that is the worry, I hope an ‘easy mode’ and a ‘pro mode’ are both considered.

1 Like

I’m not entirely sure what problem you’re trying to solve with this. Honestly why do you need to do another full redesign. Maybe I can see how batch mode is slightly easier…?

Everything else looks far less streamlined. For example, I like to take a video and make 4-5 different enhancements on it and compare them. With this UI that looks like it would be a total nightmare.

Here’s my feedback. Figure out your development process so that you release things with fewer bugs. The app is constantly broken in tons of small ways. It’s not a pleasure to use. A full GUI redesign will just introduce so many more. Figure out how to release something with a decent level of polish, then aim for a new look.

This to me is a waste of time. Maybe bring a couple of the ideas into the current interface, but a total top-down redo? Huge mistake.

Also I think I’d give more actionable feedback if you said: “these are the things we’re aiming to improve with this concept.” so we have a bit more context on how to evaluate it.

11 Likes

Would you consider a multi-monitor option (undocking) for those with more than one screen? I use three monitors for my work now and would love to use one screen to see full sized previews and another for the controls and other functions.

3 Likes

I agree with the previous speakers. A Pro mode would be very helpful. Ultimately, however, I am very satisfied with the current programme interface. It would be great to have face recognition for videos with high resolution in order to be able to upscale from FHD to UHD or a further improvement of iris medium resolution. I would also like to see the old live-view mode for rendering again, as I can only test short scenes when preselecting the filter. If I then see a bad result in a scene during the rendering progress, I can stop further progress and thus gain time.

Translated with DeepL.com (free version)

Presets are an okay starting point, but no replacement for project files.
What the GUI is currently missing is the ability to:

  • Save and load projects files.
  • Define multiple in and out markers per video.
  • Define individual AI models and settings for each set of in and out marker sections.

It does look like the batch workflow will be easier with this new design concept, but that should fall in place naturally if save and load is implemented.

Maybe the way I use TVAI is unique. I mostly do TV shows DVD to FHD. I like to cut out intros and credits. There are several instances where I would love to be able to process different models, or different settings of the same model, on the episode.
Ideally, I would be able to: define all the ins and outs, set what models and setting will be applied to each set, save a file with all those detail, be able to load part or all of the settings from that file again on any video and of course be able to copy settings from one section and paste them onto any section of any video open.

One reoccurring example of why I want this is for when new models come out that are better. Instead of needing to find all the same ins and outs again, I would just load the settings file and start changing what new models and settings get applied to what sections.

5 Likes

As a user since v2.4/2.6 I would say my favorite and most usable UI was 3.x series (ie 3.5), with 4.1 being really cramped looking on my laptop screen, due to how much space the track progress bars take up and the widening of fonts, UI elements, etc.

My main input for UI changes in future updates would be to allow users to be able to hide the right and/or bottom (export) panels in normal mode, with some refining of the UI to have a ‘smaller’ mode where text and icon spacing is massively reduced to allow wider side-by-side comparisons.

In addition, I really miss the 4X comparison with the AI models like in the older versions. The UI changing every 6 months and removing features is frustrating for finding the best model, and the inclusion of the ‘Profile 1 vs Profile 2’ UI idea doesn’t make it easier for quick enhancements or newer users (of the update, or in general).

“Pro” mode should have every UI element visible as is now, and/or more like 3.5’s themeing, less colorful aesthetic but that’s just my preference.

Standard mode should be simplifying the visible options (ie, no Preview 1/2 swapping) unless checked off in the View menu and/or in the settings, like Second Enhancement and other options are.

If someone wants a Compact mode, then the new UI, could be used there, but I’m worried it’ll introduce other issues that didn’t exist before, like how 3.x’s UI framerate drops caused my Freesync screen to stutter and flicker, while prior versions didn’t seem to cause it as badly

I also would like bugs with the window height being jumpy on some systems to be fixed, on my Windows 11 laptop with Explorerpatcher (Win10 look), V3.5 and 4.1 can’t decide exactly what the maximium window size and scaling should be, and frustratingly adjusts depending on hover over. Not sure what’s causing it or if it has to do with my DPI scaling, but it hasn’t been fixed and I haven’t seen any other program have it

One issue with this new UI would be making it ‘easy’ to lose track of where the interface’s controls are if it gets too small, like with the Filter text + button in the example video. I’m also kind of got the awareness / focus of a goldfish, and if it’s easy to accidentally close the Upscaler settings + Model popup UI, I might end up forgetting what I’ve set and second guess myself lol. Having the panel / side pane system currently lets me constantly see the thing I’m focused on and to easily make adjustments + click preview again to see if the smaller changes were useful. I’m a fan of Proteus manual because it lets me ramp up the sharpening, or reduce the decompression etc

This is looking great. The best features from previous versions, a cleaner interface, and finally, back to a quick and easy preview.

Project support should be out soon.
The scenes support with this GUI will look something like this.

3 Likes

No.
Stop with the redesigns.
Just get what exists fully working and improve the models!
You are losing your way, and people are going to stop paying.
How would you feel if you had to work in a building that was not only never finished but kept getting torn down and rebuilt before it was ever completed the first time? The top floors aren’t done, the elevator only goes up halfway and you can’t remember the last time the sinks worked… And they’re already talking about tearing it down and building a new one. How long do you think you’d continue working there?
Your development cycle and where it places value and priority is ridiculous.
I still haven’t moved to v4 yet because it still isn’t fully operational.

19 Likes

I am not sure this is helpful for me. More attention on the CLI version would be good or performance.
I am using this in a workflow. I need to be able to trim and crop easily and use time codes (or frame numbers). This looks more like a Final Cut Pro Direction.
Things I need:

  • Entry of Timecode or frame number to start and end a trim.
  • Easy crop functions
  • CLI
  • Ability to add chapter markers for MP4 files. I’d rather be able to scroll through the video drop a new markers with text and be done with this.
  • ability to export a working CLI command. (The CLI reference did not work.)
  • Previews from the beginning are not as useful as from some point in the middle
  • MPEGSTreamClip now defunct used to have the ability to fix timecode issues and put more than one trim (in/out) marker.
  • Ability to analyze and recommend filters for best results. I really don’t want to try all of the combinations and permutations to find the perfect setting.
  • Export for specific devices (iPAD, iPhone, TV, Web, and Distribution) with proper frame rates would be a nice thing.

The UI isn’t going to sell the product. The work flow is what will sell the product.

I really want the editing to be simple. I have Final Cut Pro and Adobe Premere for advanced editing. Trying to make this more for production and I have thousands of videos to enhance.

Overly simple UIs don’t really help.

6 Likes

Looking good temp processing at least Main10 Profile.
Would be good for program to have more color depth.
Snapshots could be Tiff frames or 32Bit exr files.
GUI could look at frames too.
For workflows export to Final Cut Pro,Premier Pro,Vegas ,Divinchi Resolve Studio,ect.

You have to define and prioritize use cases and then find solutions for those use cases. I am concerned you are working backwards, throwing out concepts and trying to make them fit in our workflows.

I started thinking about what would be my 3 primary use cases (or user stories):

  1. while navigate a video, allow to find the best model and parameters with A MINIMUM NUMBER OF CLICKS AND TIME WASTED. Pick a frame, navigate easily, apply the model IN REAL TIME on that frame (and subsequent frames). I need to see the original frame and the enhanced frames, if you want to compare 2 models (A/B), I still want to see the original frame
  2. when the parameters are chosen (models, parameters, cropping, begin, end, output format, quality…), process the video.
  3. easily retrieve previously set of model/parameters to apply them to a video

I am sure there are many others use cases and priorities, e.g. I didn’t get into execution of multiple models in sequence. Maybe setup a way for us to express and prioritize requirements if you want, and only then, from those prioritized requirements develop UI concepts.

Thank you!

Now commenting on the video itself.

  1. please avoid drop-downs menu, those generate too many clicks. I would prefer to have selections on a side panel, the frame does not have to take the majority of the screen real-estate. We can always zoom in to see details
  2. will the processing apply on the selected frame (s) and will it update in real time? Processing ONE frame should not take much time
  3. will I be able to see the original and the process frame side by side?
  4. when I navigate from one frame to the next, will both frame sets be in sync?

Thank you.

4 Likes

This comes across as a bit insulting, but it’s fairly accurate.

Please just ask us what we want.
Do a poll of the Top 3 of Topaz’s ideas, and the Top 3 of users’ ideas.