Feature Requests

It would be nice to have the ability to write on two lines or more. It would create the possibility, for instance, to write a single word on two lines. For graphic funny creation.

while using Firefly, I find my self wanting to mark or save some aspects of the image for later review. It would be great ive queries were saved or tied to the image before submitting to the galley. on a similar thought. if you could save a selection of a image to allow that part to be used in the next step (prompt or style change. IE: creating a robot in first wave - selecting the feet and head to be used but re generateing the rest of the image. 

[EN] Since you have an artificial intelligence that can create an image with text. You can have an artificial intelligence that defines the different elements of an image. You could have a detection of the different elements of a scene or a video. List the elements that make noise with a timecode. In this way, it can facilitate the work of sound designers.   Once you have an AI that allows you to create sounds according to a text, you can link the list of elements that make noise, their timecode, their position in the image and thus create the sound design of a scene or a video. [FR] Puisque vous avez une intelligence artificielle qui peut créer une image avec du texte. Vous pouvez avoir une intelligence artificielle qui défini les différents éléments d'une image. On pourrait envisager une détection des différents éléments d'une scène ou d'une vidéo. Faire la liste des éléments qui font du bruits avec un timecode. Ainsi, cela peut faciliter le travail des sound designers.   U

New users will undoubtedly get overwhelmed with options - A page where a baseline directory can be created that gives a very quick visual reference to what type of image or style can be expected in an output. I understand there is a styles tab within Firefly now, but a standalone directory would be great to browse through for inspiration without being distracted by all of the other UI elements.

To make branded images, when this is out of beta, making sure the colors generated match company style guides would be incredibly powerful.

Text-to-image in edit mode:I noticed that the thumbnails of each image style featuring the flying balloon does not effectively showcase the effects. Scanning them all together can be challenging. It may be more beneficial to use a carefully selected, generated image that demonstrates the best the effect.I just find it difficult to discern the differences between them due to their small size.

It takes a bit of getting used to going to the bottom of the screen to type in the text for the image you want to generate. Took me a while to find it on the first go which isn't a great UX. It should be at the top or right in the middle.

ability to upload multiple images and combine them to one image. Bonus if you can pick which aspects of the images to include.    one application: a cyber punk style image and a selfie. You can tell it to pull the cyber punk style and color scheme from one photo and apply to the image of the other photo. "and also, throw in the 80's looking grid in the background"   another application: you have three shots from the same shoot, same take. You like the person's smile from one, the stance from the other, and the lighting from the last. you upload them all and get the best of each.  

It is suggested that in the function of generating pictures from text, add the function of reverse prompt words.

Saw this in one of the videos, but wanted to make sure it was captured and voted on here. Would be great to change the weather in an environmental image in the same way we can currently change the lighting or composition with tags.

It's quite common to get an image you want to use in a painting or as a texture in a 3d object that has too low a resolution. So would love to see Adobe Ai handle this too. There's a number of pieces of software that does this already such as Topaz Gigapixel AI, but would be great to have it built in directly into firefly. 

If you're going to limit Veo generations, at least give us other reliable options. Please put the other video generation models in the backburner if they cannot be reliable in this day and age.

I want to highlight an area say KEEP or REDO and put in another prompt to make refinement edits 

I have created a project for myself that started with Firefly images further edited in Photoshop.How and where should I share it?

Is it possible to select a result you like and then start from that result to fine-tune it, instead of clicking Refeesh over and over to generate more random results?

There should be an option to compose or redo from an image.

The text filter is flagging benign prompts as having offensive content. The challenge users face when presented with the warning flag from the content filter is that it doesn't tell me which word, or phrase, is causing the issue. As a result, finding and removing the offending content is challenging. I usually just wipe the entire prompt and try something else because I have no idea which word it could be(I never use offensive content). Additionally, there is no way to dispute the flag, which removes important words from the english language from being used in prompt craft.  My suggestions are as follows: If content is flagged as being against some adobe policy, provide visual feedback so the user can quickly identify and fix the issue. Some examples of how you could do this are: change the text color, underline the text, provide a list of words and the reason why they are banned with a button to file a dispute. The format would be:  Badword: This word is

I think it would be wonderful if it was possible to see the minor changes from different prompts and choices in a list. A bit more like mid-journey, if I am being honest. When I pick "show similar," I don't want the image I liked to disappear, or the other ones. It might be nice to rank all four images that appear to help generate similar but more specific to my exact needs. Maybe it would be useful if I could make a lightbox of the images that I like. Also, whenever I select the "reference image," it goes in a totally different direction than I expected or hoped. Thank you for your time!

Sometimes the text to image generates some pictures that deliveres exactly something that I visualized in my mind - but in seperate pictures. It would be really awesome if I  for example could select "Background" from picture A and figure from picture B and combine them.

I how u guys can create a image to vector conversorsomeplace where i symple upload a image and transform it into a vector

mixing two images for examplea computer 3d model of a buildingand a particular style of a architect or architecture movement like brutalismwhich can produce blended result

I hope you take a look to some apps that give you examples to click on it to add prompts like imi prompt that would be a great idea 

It would be wonderful to have the ability to choose one primary image from a group of four and then have the flexibility to make modifications to that particular image. These modifications could include adjustments to color, focus, blur, changing the person or object, or altering background elements, etc. This process would generate a new set of results and include the original image selected. For instance, in the image provided, the first dog is perfect, but it would be great to change only the color of its fur or the background.

It would be really handy to be able to add a picture in the text to image Ai. This would help narrowing down the search

One of the most frustrating things about where a LOT of companies are going is to subscription based applications. I have a Mac Studio Ultra, M2 Ultra CPU, 128 GB Ram, Screaming fast SSDs...Yet, everyone wants to run their software in the 'cloud' (Clouds are not fuffy wonderful places, they are Sub-Zero, freezing miserable places, that it is very hard to get to, hard to stay there, expensive to get to.. and you don't own them... and they can blow away on a whim.  Which is why cloud computing is so well named...) I don't currently 'rent' software, so I stopped using Photoshop, and now use Pixamator and other applications that I can purchase, and run on my kicking fast hardware.  Something that *I* have control of, and if it stops working it would be because I upgraded the software/os and I can always go back to a backup using Time machine. I also have 4 TB of space to store the software, so no problems there.  Paying $50/month to use would price this out of