P: Maintaining a scene or character across different prompts

[James2808541164de]

Let's say I wanted to storyboard a script and produce different angles of the same thing. It would be nice to have something that holds on to the core look and objects within the frame so we can get a wide, medium , and close shot of the same scene.

[Tracy.H]

Thanks for suggesting! I think we have a few different ideas that might help with your use case. Have you checked out this video? (Merging all similar threads here so we can track the popularity of this feature request)

[Afro29114102tsnh]

Hello, Is there any possibility to keep a specific prompt art, like if a person creates a character, to be able to use that character in other settings or in other prompts without changing the appearance of the character and maintaining the first prompt.

[defaultb3jsoit5wpvm]

This would be an amazing idea, to include a setting for creating storyboards and going even futher create basic animatics for pitches.

You can use the tool as it is for this to a point but a specialised setting for this with functions just for storyboards, would make creating fast storyboards. Not every project and client has the budget to create full storyboard concepts.

[Charles29191999ljm8]

It would be useful for sequential artists if there was a way to train characters to be used in storytelling and storyboarding for cartoons and films. A poseable dummy like with controlnet would be beneficial as well, with a way to created with layers instead of out and in-painting.

[StarFinesmith]

Yes, Charles, I completely agree! Storytelling is one of the main things I do. Being able to manage the AI characters for consistency in stories would be most helpful. 

[Henrik Heigl]

Hi,

for that Adobe has also the Product Character Animator https://www.adobe.com/products/character-animator.html 

So, in Adobe Firefly you can build your Character and then in Adobe Character Animator you can make it pose.
Hope that helps.

[Shon-Lueiss5FB2]

Adobe Firefly is uniquely capable at creating a polished subject. As a writer, I found the depictions of human subjects more accurately captured what I described and with less effort than other generators like Midjourney. I also found the text-to-image generator was capable of producing imaginative logo treatments (though it has issues keeping the text in order). 

A new feature I recommend is the ability to ask Adobe Firefly to "remember" a subject. For example, if I create a logo treatment or character, I want to be able to name them and use them in future prompts. This would allow creators to bring a consistent subject across multiple environments, settings, etc. 

Another generator called Runway is currently offering this feature. I am also experimenting with its use to see how successful the model can pick up this ability. I believe getting to this feature early would give Adobe Firefly a significant advantage.

[Tracy.H]

@Shon-Lueiss5FB2 Merging your thread here so we can track the demand for certain features.

[Michael25883953lcm2]

When I'm setting up something more complex (such as with a background and a foreground and maybe a character or two) I wish I could lock the results of a certain element that I really like while still changing the other parts. I had one come up yesterday where the background was AMAZING, but the character was off. Clicking the "more like this" option was painful as the background disappeared forever.

Merging your thread here so we can track the demand for certain features.

[AndrewBuckle]

When you click generate, it would be great if you could select an image with a face or a car etc and then make it always use that face and re-generate with different scenarios etc, so you could then build up a selection of images with the same person. At the moment, hopeless for a comic book or any number of illustrations as the face keeps changing on each generate - certainly not useful if you are going to create a video from it as the face would be changing all the time (though that would make for an interesting abstract video)

Reason - for create panels for comic books etc, no use if the face or the car model etc keeps changing

Merging your thread here so we can track the demand for certain features.

[36 Styles]

A Masking method the same way done in InvokeAI, where you can brush the parts of the image you want to change or invert it and effect/change everything that has not been painted. This would be incredible to see. 
https://youtu.be/OFiJ1Bv0FIM

[Mark S. Johnson]

Firefly is off to a beautiful beginning.  Nice work, team!

I've been working with Midjourney and Stable Diffusion for awhile and one of the biggest challenges with them is creating a consistent character (or characters) that can be used in multiple scenes with consistent clothing, yet altered poses and expressions.  If Adobe can beat those guys to this capability, it will be absolutely game changing.  Whoever figures this out first is going to have to beat storytellers like me away.  🙂

Thanks for considering this idea!

Merging your thread here so we can track the demand for certain features.

[John29654733jgfq]

Firefly is pretty good for what it is, but currently, the program is of limited use to me and is a bit like playing a slot machine. I need to do mockups for a stage performance, putting multiple actors together to work out blocking, lighting, and to give the producers concept photos of what we are attempting. Each image of what should be the same character is generated randomly, so there is no continuity between photos. I realize that the AI data set is limited to Adobe Stock, but even historical figures like Mark Twain are beyond the capacity as it sits. Even common and unchanging elements like a full moon render as abstract even in photo mode. I have historical photos that I am using for characters( photo A), and ideally, having an AI to help position them, change their outfits, or alter their location, would be a huge time saver and worth the price of admission. If there is any way to input our own image data set, then let the AI work from that (i.e., Take this person from photo A and put it in the pose from photo B), it would be easy then to cut and paste them together to work out scenes. Perhaps as a function of Photoshop? Thanks for the chance to experiment. 

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[rafałw53945353]

I think this is nice option if you have photo but you need more shoots from situation on photo. Like the same model but different position, different background and light

Merging your thread here so we can track the demand for certain features.

[MonstersAndUnicorns]

I want to develop a whole series of images with one created image. The style and the figures, for example, should keep the same character, but the scenery should adapt. E.g. a cat sits on a chair (style was chosen) in the next picture the same cat should sit on the bed and so on.

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[Michael30076736p5eb]

an app that can take a charater from an image and use it in other images. That way you can have a reoccuring character for a comic book or a series of images that could show the character doing other things showing other angles etc. From photoreal to cartoon, using the same style if desired

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[Carlos Caicedo]

I agree I would find it usefull that once I make an Ai character or import a 3D character have the AI place my character in different scenes and lighting, clothes and have the chracter be the same recognisable character regardles of the AI scene. The scene woyld also need to to be consistent but AI generated

[B30050314wsn1]

Totally agree!

[Abb26696536ugs8]

dear team,

I'm trying to create a storyline for a childrens book and the subject (teddy bear) keeps changing when I create different pictures.

It would be amazing if it is possible to holde the subject still when I'm generating different pictures.

Many thanks,

Abb

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[cory295935243f11]

IT would be great once you find an image to build upon that imiage. IF you create a character, you can save that character or image and give Firefly more commands building images with a specific character or idea being the same in every image just with different backgrounds or situations 

[CSI Productions]

I'm considering using FF (when released) to generate imagery for a children's book my daughter wrote. I've played with the bets version to that end and the results are promising. The one issue I'm having is getting a consistent character in the various scenes.

As great as the look and feel is, as I go through the story and generate tests, I don't see a way to keep the same character from scene to scene.

Possibly a feature that lets you identify a character (maybe circling the character and giving it a name) then include that name in future prompts.

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[Noah30243189yy0c]

Someone in my industry was telling me they use firefly to generate storyboards. Adobe should add this function. You could generate main characters that remain throughout the storyboard, attach quotes to the storyboard for export and presenation, and include more camera movement and action arrows throughout.

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[N.K nikhil30295365ix8n]

i want more pics with same face in Adobe Firefly

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.

[N.K nikhil30295365ix8n]

i want more pics with same face in Adobe Firefly. it's posible or not ?

[Ganesh30345632ng6g]

The ai not able to convert our toon avatar with face same

Merging your thread here so we can track the demand for certain features. You can upvote the original post to let us know you want this feature.