Feature Requests

Firefly needs upsacale option to evolve the option you like, add details, and even better to make changes on it.So far Firefly has an enormous quantity of distortion just look at the birds below. also poor quality in general.Here is a vs Midjourney vs Firefly, utilizing the same prompt "cabin in the woods surrounded by trees with a creek nearby Make it middle of the summer birds flying high bright ultra realistic 4K"The biggest problem that Dall-e and Midjourney have is the unavility to make changes. I think Upscale and changes would be definitly a game changer for Firefly. I love Adobe, I use several Adobe Apps every single day and I want this to be the company I chooste to create AI images, but you have to run, the race already started long ago.

To be able to send the image to a desired design tool, this would make working with Adobe Firefly more seamless. Or even a step further would be to also integrate this into each tool where can call it up as needed.

Delighted to finally test the beta. First suggestion: Change the rating system from the simple thumbs-up/-down to a one-to-four-star system or something similar; I'd like a bit more nuance than a strict up or down vote. For example, in my first prompt, I asked for a close-up of a human hand, with all fingers extended. As expected, all results were weirdly formed, and one had seven fingers. They were all "wrong" (albeit amusing, as I expect this to happen with all image generators), but the one with seven fingers is more "wrong" than the others. I'd like to be able to express that in a quick rating.  Of course, this certainly is not a deal breaker; otherwise, the whole system is very impressive. More commentary to come.

Create a new kind of file extension for AI-generated images to reduce the likelihood of facial images in the Internet.

Currently, some options cause an automatic Refresh/Regenerate, such as moving the Strength slider for the reference image or choosing a different style. This causes unexpected additional requests that take up time and compute resources. Please default to only refreshing when the actual button is pressed to avoid delays and wasting resources.

I would love to see vector generation in Firefly. The text-to-image model is already amazing, and I think the ability to generate clean vector graphics via prompt will be an extremely useful tool!

when changing the type from  I.E. art to photo, make it so it does not change the results displayed at the moment.

Maybe we cold change the background color and opacity for text effects?   What do you guys think?

would be great to be able to automatically apply a generated image as background or sky in a picture from lightroom

While clicking show similar there should be an option to edit our prompt for obtaining similar images. This is necessary as the results are random and we may not be able to get that same image again. This will also help to improve background without changing foreground.

Adobe Firelfly's text effects are pretty nifty, but they all seem to keep the hard border of the text. I tried to do a smoky text with loose borders, for instance, but there was still a hard black line where the text ended. It was like the text was written in black with smoke superimposed over it (which is actually probably all it is), occasionally extending beyond the borders. I would like for that hard border to be completely absent and only have the edge of the text defined by the effect itself, if that makes sense, making the text look like it was made of smoke rather than text with smoke in it.

In firefly muss dringend das Text Limit in den Prompt-Eingabefeld größer oder aufgehoben werden. Auch wichtig , dass der Bild Verlauf im Hauptfenster gelöscht wird nach dem Beenden oder Schließen des Browsers und wenn Bilder gelöscht wurden. Kling AI bitte auf Version 3 anbieten und mit der Möglichkeit Motion Control und Lip Sync anzubieten

It's probably very obvious but I was thinking drop shadows and 3D options for Text Effects in Firefly would make nice additions to this already very powerful feature.

Just a suggestion how about ading all these features in software like Photoshop and premiere pro where if you need an image or a new font or even a template you can do it from there i know it needs to have internet so let it be like a button it's grayed out when there is no internet and is darker when Internet is there you can access all ai tools like image creation just like you would in a browser but faster and simpler 

In boards, it always defaults to nano banana in the initial generate. It would be great if there was a preference to one of the others such as ‘Firefly 4’  I never want to use nano and I have clicked it a few times before realising that and then going to the model and resetting it to ‘firefly 4’ 

A user feedback tool - like pointer / brush / arrow / circle -  could be very handy / productive to reach desired results quickly -- Sometimes most of the image is on the right track for a desired text to image result, but a certain item is introduced that changes all further results - if one could point out "Good" and "Bad" or "Ideal" and "Unwanted" elements, it could help. In the example below a few unexpected results came up for "CAD drawing luxury sedan wireframe concept"  1. The style filter "shot from above" wasn't applied - it might be productive if one could flag?2. The figure was not specified, and it could be good if it could be identified for omission3. The figure had a distorted area with the second overlapping arm4. The proportion of the vehicle seems unexpected and the back section's colour different.

Can you work this feature into Express so I can incorporate the idea into different projects?

menus does not have background. is transparent tex, and  opens up to instea dto the side..been able to chose the seed of the render will be great

I am writing in Japanese using the automatic translation function. I've only used it a little, but I'll write what I felt. 1. I was impressed that it already supports Japanese input.Unfortunately, there are no Japanese-compatible fonts in the options, so all the characters are thin. I think it would be perfect to be able to select a font that supports Japanese.2. Characters may overlap and be displayed. I thought it would be convenient if I could download it as a PSD layered for each character. This is not a high priority because it can be solved by generating each character.   --------------------------------------- 日本語で自動翻訳機能を使って書いています。 少し使っただけですが感じたことを書かせていただきます。 1.既に日本語の入力にに対応していて感動しました。ですが残念なことに、選択肢の中に日本語対応のフォントがないために、全て細い文字になってしまいます。日本語に対応したフォントを選択できると完ぺきだと思います。2.文字と文字が重なって表示されてしまう場合があります。一文字ごとにレイヤー分けされたPSDとしてダウンロードできたりすると便利だと思いました。これに関しては一文字ごとにジェネレートすれば解決可能なので優先順位は高くはありません。

When changing styles, the base picture is almost the same. But when the prompt is changed, the picture changes a lot. It would be nice to be able to return to the picture we kept when we think the previous base picture was better.

There should be a way to get back to the reference image and be able to use it again or even just download it again. 

Need to be able to remove watermark. Users owns responsiblity for legality right to the image.

I would like my characters to be drawn in different poses and out of front using AI.I hope that such a functionality will be provided in Firefly.

Give the ability to tweak the strength of the effect.  Almost like a wallpaper on your computer do you want it to tile, or stretch to fill kind of thing.  I want a wood plank texture not a bunch of small pieces of wood.  Good start though. 

currently on “text to effect” it is producing same effect if the character is same as examples below… if the every single character on the word (even if the same character repeats) produces a unique result, it will be more organic to view…