Feature Requests

Current behaviour: 1. Use lasso tool, release mouse button -> floating toolbar appears. 2. Hold shift to select more -> floating toolbar is not hidden until mousedown.   Better: 1. Use lasso tool, release mouse button -> floating toolbar appears. 2. Hold shift to select more -> floating toolbar hides on shift down*    * unless toolbar is mouse-hovered and there is a possible scenario when shift+click does something different on the toolbar itself.   I know the toolbar can be pinned and hidden manually, but this would be great for when you keep it floating. The toolbar itself otherwise gets in the way a lot.

It would be nice if Gen Fill:   Idea: If I Gen Fill a new "mouth" or a new dress on someone, if Gen Fill could create an Asset Number for that particular fill and if I go to the next image of the same model, I can tell Gen Fill to create a mouth/dress  using Asset Number.   This would work good so you can change a series of shots into the new mouth/dress and not be pot-luck on every sucessive photo.  

Testing out the new Photoshop Beta (24.7), emphasis in my opinion should be prioritized on easy and accurate smart AI masking. Also, the removal tool is nice, but we should be able to select a sample of the image and prompt AI to "remove fence" and it should remove the entire fencing. Or AI to remove repeatable patterns in an image. AI Generative Fill is nice, but AI Masking and AI Pattern or AI patch removal is needed to make our lioves easier and to save incredible time from making difficult maskings. 

The generated results from the new Ai fills are often unseable (reselution, details)

Setup   Photoshop on the web (Beta) Version 202206.12.0 macOS Monterey 12.4 (MacBook Air, M1 2020) Microsoft EdgeVersion 103.0.1264.51 (Official Build) (arm64) Google ChromeVersion 103.0.5060.114 (Official Build) (arm64)   Description of Problem     The word "toggle" is used in all the tooltips for the four action buttons in the Taskbar:   Toggle layers Toggle layer properties Toggle comments Toggle info   The Spectrum Design System explicity says the word "toggle" should be avoided.   Furthermore, as these are icon-only action buttons, the tooltip for each button should be the text that would have been used for the button's label (as per the advice in the Spectrum Design System guidelines).   Suggestion   Rewrite the tooltips so they become simply:   Layers Properties Comments Document info   This has the following benefits:   It is quicker to read what the function is for each button is. The most important word

Could you please swap the color tones of 'visible layer' eye icons? - make visible a BLACK icon, - the hidden layer icon GREY My mind automatically thinks the darker 'bolder' layer should be visible - but this isn't the case... it's the lighter grey eye icon that is currently used for the visible layer...  ... doens't seem to be intuative in it's current state - and always confuses me...

I'm a retoucher and been having a play with using Generative Fill in Photoshop Beta.   It works great with tedious tasks like removing shadows, reflections, background objects, etc without having to do much masking. If the AI layer mask is good, adjustment layers clipped to it and/or you do some manual work on top to make it more believable.   The issue is, if I have any adjustment layers on the overall image, the AI will sample EVERYTHING essentially doubling up on the adjustments instead of sampling just the selected layer.   Left to right: original image, Generative Fill without any adjustment layers and Generative Fill if you use it with adjustment layers applied (I made my curves and hue/sat more dramatic and my mask really rough to illustrate this)   P.S. Yes, if you: do it before adding adjustments/do all the adjustments in Lightroom/do it as a final step, then it's fine, but when using a non destructive retouching method, makes things frustrating

Would be nice if the generated background was behind the subject with mask, so we can adjust the mask. Similar to neural filters. Right now if the background overlapped with the hand, you can only copy the original subject on top. Would be nice if I can modify the mask to restore some of the original subject if needed.

Generative Fill is absolutely stunning.   While it is cool to have 3 different options to toggle between after generation, I wonder if that requires additional processing power compared to one. I would find it helpful to have a preference setting where I could switch it to 1. Reason being that the majority of the time, I'm happy with the first generation, and don't bother to look at the alternatives. However I would want it as a setting in case I want to try something a text prompt more creative then un-cropping.   Additionally I would love if I could have an option for it to be just one button press. Maybe I can set a keyboard shortcut, I haven't looked into it yet. Most of the time I don't enter text, I just want to un-crop something.   Finally, I did get a content warning that appeared, for no known reason. I was confused but the pop up window automatically disappeared before I could read it. I think it was a bug in the interpretation because I didn't generate anythin

I find that certain prompts are used repeatedly. (hyper-realistic, high detail, bird's eye angle, sunset,)  I wish photoshop could simply list the last 10-20 prompts that I entered and offer a check box next to each prompt so I can select the promtpts I'd like to include before I click "Generate" again.  Reducing the amount of typed input required to gernerate a new fill.

       hi！Recently, I tried the new version of photoshop beta AI. The function is very easy to use. I have an idea. I don’t know if there will be this function in future versions. For example: open a road picture in the software, open a car picture, and insert the car picture into The road picture, and then through AI calculation, the car picture can be quickly integrated into the road picture, which can solve the car pattern you want very well.   Thanks  

Ok, generative fill still have a lot of flaws but as a first iteration it is quite impressive. However, I think one of the most important things to improve is "consistency". For example: in an album with multiple photos of the same subject in the same scene, I'd like to add "angel wings" to the human subject, and I can get a pretty good looking wings that doesn't that doesn't look too "off", however, the problem is that it is pretty hard to get the exact same type of wings to add to the same subject in a different photo of the same album because when I use generative fill to add wing to the subject, and different wings is created.   Excuse if my explaination is a bit hard to understand, English is not my first language.

I would like the same control over the generated image as I have when using the web version of Firefly.

Let's say I want to use this image below of alphabet soup, but have it spell out a custom phrase. Below is an example where I selected the middle portion of the soup and using Generative Fill told the AI to generate the phrase "Bar Mitzvah". You can see that it struggles with building the phrase correctly, but understood the idea that it had to use "noodle letters" to spell it out. Because the AI was able to understand where to pull the type texture from and how to rebuild it, my thought it that it could use this information but then apply it to live type. So once you approve the AI output for how the type should look you can then create a type box and input your phrase so that the words are correct, and you can move them around in real time for placement.Beyond this, I could then see selecting a font so that you have the aphabet soup noodle texure, a live type box with your exact phrase, and you can then tell it to keep the texture and placement but make it cursive, or serif for exa

Ability to choose between a transparent image generated by AI or a regular image with a background. This gives you more flexibility to manipulate the image by adding effects such as shadows, highlights, and reflections. Additionally, you have the option to use your own library of photos to integrate into the scene. In addition, the ability to create multiple scenes from the original chosen/source-generated image. Furthermore, the generated pixel resolution is larger than before, exceeding 1024x1024px with IMPROVEMENTS for Generating HUMAN figures, Hands face body, etc...  PLEASE VOTE AND MAKE THIS HAPPEN ALREADY!

It's a great tool, but please please please impliment at minimum browser standard input support, meaning:   1. Spell check. Very important. 2. Keyboard control. When I am in generate prompt field and I press ctrl+left or right arrow I am attempting to select a word, not move my mask and clear my prompt. 3. Proper undo/redo would be nice 4. Somewhat unrelated and potentially not inline with the "workflow" style here, but if I am trying something and I accidentally click away the propmpt is gone, can't get it back. 

Put variations according to the result like mindjourney. For example, I generated a wooden barrel, there was a suggestion that I liked but I couldn't generate a new one from it, I had to recreate a new generation.

In this day and age is there any need to cycle between the two tools? 🙂 Can we just use alt when using the spot healing brush to turn it into the healing brush?  

Is their any animation tools at all in photoshop web? the photoshop knockoffs can animate in a knockoff photoshop fasion, but I pefer to use photoshop web. Is the timeline tool or any animation tools within the web app?

We’re excited to hear your ideas for Photoshop beta and look forward to seeing your new feature requests! Note:  For feature requests for Photoshop on desktop, Web, iPhone or iPad: Head over to the Photoshop ecosystem ideas board.   You may ask, What you can do to increase the chances of being heard?    One request per thread: When creating an Idea or Feature Request thread, please keep requests for features to ONE request per thread. Doing this will allow others to vote on the single, specific feature and not confuse Upvotes amongst a list of feature requests.   If an Idea thread has multiple requests in it, you may be asked by a moderator to split the multiple requests up into multiple threads so others can vote.      Example Idea thread:     "When using the brush tool, it would be great if it was pressure sensitive when using my stylus" "I really need X tool, when will that be implemented?" Additionally: Any extra detail you can