Feature Requests

Firefly and its Partner Models should allow creation of Action Fight and Horror movies with relaxed prompt restrictions. Currently it is very difficult if not imposible to create such short movies due to prompt input and output restrictions.

Firefly could easily address a direct PNG downloads:Offering PNG downloads directly, particularly for images requiring transparency or high detail.Providing better resolution options for export, so users don’t have to rely on external upscaling.

I would like Firefly video to allow subsequent output of content to be consistent in look and feel with, so new content with clearly specified (by the user) variations. This is possible to a degree but has limitations as I'll cover. Examples of what I've tried to do with limited success.* Create a powerful looking character by the camera angle and the expression in the character. Then have the same character looking vulnerable. * Create character in video doing one thing for a few seconds, then the same character doing something different for a few seconds.   In the current beta version, changing the prompt slightly can give significantly different looking results.You can use a reference image but it is restrictive. As mentioned in my post Firefly Video 'Shot size' and 'Camera angle' with reference image, if you use a reference image, you can't change the camera angle by the settings or in the prompt.   Firefly Video has a seed number setting. My understanding of a

can we have ADJUSTMENT PROMPS?  for example:keep the bustop but the guy's bike should be a mountsain bike OR keep that scene but change it to a rainy day .....   often ajusting the original prompt text inytoduces new random results and changes the image that is 90% correct.

It would be useful if Firefly video can understand how visualise a character talking beyond randon mouth movements. I have tried something like:<character> saying "hello" Firefly kindof understands it with a human character, showing the character waving a hand, no mouth movement. With an animal, there is no indication of it greeting the viewer.When I use "taking" instead of "saying <something>", I get mixed results, sometimes mouth moving, mostly nothing to indicate speech.

I love some of the autofill images especially after a few itereations. However, I have no way to go back to the image I am referencing or even a way to save that new image into favorites folders 

Firefly image and video doesn't communicate how it is dealing with contradictions in what is specified. Example 1: In Firefly video in the 'shot size' setting, I choose 'Extreme long shot' but in the prompt text I include 'Extreme close up'.  The result after multiple attempts was that the prompt 'Extreme close up' took priority in the output.  Example 2:In Firefly video, my prompt had "red and orange koala, grayscale" (abbreviated version). Results with multiple attempts included simply a grayscale koala, a mostly grayscale koala with some hint of red/orange in its face and ears, a face that starts as mostly grayscale that animates into a more orange face, a face that is orange but ears, eyes, nose, body are grayscale.The above are clear contradictions. There may be more subtle ones that a user does not realise. A way Firefly could deal with things like the above, both for images and video),is to show the user the actual prompt that was used - so a new Firefly version o

Hello,"I had an idea while editing my photos. I used an AI-generated background for one picture and loved it. To maintain consistency, I wish there was a feature to apply the same AI-generated background to multiple photos, making them look cohesive and natural. Would be awesome if you could consider this feature!"

Hallo zusammen, durch das benutzen ist mir aufgefallen das die Funktion "Bildformat" manchmal bis garnicht geht. Da man soviele verschiedenen Formate hat und diese auch unterschiedlich braucht, wuerde eine funktion fuer die Formatvorlage echt super sein.  Warum, da man immer nur ein Format bekommt und die Hochskalierung nicht immer bei mir Funktioniert, was vielleicht auch bei anderen vorkommen kann. Mfg Simon.

Posso não estar achando, mas poderia ter um local onde ficasse salvo as imagens mais recentes criadas. Pois todas vez que fecho o FireFly, aquilo que eu já criei, imagens e o prompt, todos eles somem. Se caso eu quiser continuar depois usando os mesmo prompt eu não posso, pois tudo foi apagado. Então, uma biblioteca onde fique salvo pelo menos as cirações mais recentes, por exemplo dos ultimos 3 dias, já ajudaria nesse ponto.Repito, pode ser que já tenha isso lá, mas eu não estou achando, então nesse caso se já tiver, seria legal ficar de uma forma mais fácil de encontrar essa biblioteca.

I'm currently working on a customer journey for a presentation and would love to use Firefly to illustrate various product use cases. However, I'm finding it challenging because the AI's current capabilities are somewhat limited in terms of creativity. One issue I've noticed is that when I specify a prompt with the name of an object, the AI only generates it in a standard upright position. For example, even when I enter a prompt like "an overturned chair," it only produces a chair with all four legs neatly on the ground. To create more dynamic scenes, it's important for both people and objects to exhibit movement. It would be great if an update could enable the AI to understand not only the relationship between words and objects but also the posture and context in which these objects are placed. In the future, I hope to be able to input prompts like "a man who climbed onto a chair to reach a can on a high shelf and accidentally fell" and get images of a man sitting on th

Since you can upscale AI image renderings from firefly, the tech is already there. The idea is simply allow for the user to upscale an image they upload. Photoshop and lightroom seem less capable for this. Sometimes we need to have higher quality versions of images we have created in the past.

It would be good to have the option to save as a JPG, if possible

Firefly is in need of a context buffer that will allow the continuation of a prompt without having to inject a modified full prompt every time. This is completely doable in LLM architectures, a routine in text generators like ChatGPT, Copilot, or Llama, and it ties in with the repeatability problem, whereby we can't generate multiple images of the same house/car/character, but are always presented with a random variation on the theme.   Firefly would become genuinely useful if it could retain an image and execute minor edits on it via text input.   As an example:"A cartoon mouse with black ears and nose wearing white gloves and red pants." [Firefly generates an image] "Make the ears rounder." {Firefly modifies the selected image.] "How about blue pants?" etc.

Me ha sorprendido su generador de imágenes, he visto como lo han ido mejorando con el pasar del tiempo. Pero quería que ustedes se enfoquen en hacer otro producto, enfocado en la generación de video, mediante un prompt, producir un video, y que también exista la posibilidad de subir una imagen de referencia (inicial y final), para producir un video, eso sería fantástico, creo que lo pueden lograr, ya que tienen bases de inteligencia artificial, gracias.

Ich habe ein für mich passendes Bild generiert. Zu einem späteren Zeitpunkt wollte ich dieses verwenden, jedoch hatte ich das falsche Format gewählt. Eine Funktion, die die bereits generierten Bilder zunächst sichtbar macht und im Anschluss im FGormat ändern lässt wobei der Seed gleich bleibt wäre hilfreich. 

I have literally thousands of photos and graphics that I have personally created over my 30 years of art and design. How can I use Adobe AI to reference those images to create something new? I have been a paid subscriber for decades. I have posted many images on Adobe Stock, but I would really like to choose a series of my own images and have AI in Photoshop, Firefly, or a new App to generate some new ideas on how to combine them.thanks, Mike

In the "Edit -> Insert" mode, please add the option to insert a reference image as well, not only a text prompt —so that we can specify: "Insert this specific image at this specific location."

No where in my adobe account can I find a detailed report of all the generative credits used… I have to “trust” the count that Adobe gives us with no way to verify any of it except to manually track it monthly.Any time Firefly or any 3rd party model errors out I am pretty sure I lose what ever credits it said I used but I wanted to see what was really happening but there is no way to see the details.  

Sus competidores generan rostros, cuerpos y expresiones con mayor precisión y necesito convencer a mi jefa que Adobe es mejor!

Once i go into generative Fill, try out stuff and decide, that i want to reengineer my original prompt and i click "back" in the left hand upper corner, it links me to the Firefly landing page and i have to rewrite ALL of my prompt again. Very annoying, there was a version where this didn't happen.Pls fix it, or open a second tab once you go into generative fill or something, just ANYthing so you don't lose ur original prompt page.GreetsJ

It should be nice when i'm able to generate my own pictures (which i can upload) in the style of another picture of my own.

Hello, it would be great if we could manually enter aspect ratios ourselves. For example, aspect ratios like 1:2 or 2:3.

- Pouvoir créer des personnages récurrents - Pouvoir choisir le niveau de qualité de l'image- Garder ses favoris avec son compte, quelque soit le poste de travail utilisé- Pouvoir importer une photo et la transformer en dessin, voire la mixer avec une autre image

Generating an image from a sample uploaded by users to the AI. What this means is that a user might want to replicate a particular image but also wants some or all of the texts on the image to be gone as it is being generated by AI. Or maybe adding a little bit of alterations to the uploaded image.