Any Advice/Ideas on Creating Products Based on Famous Painters' Styles?

[Artsiren]

Hi All,

I'm thinking of creating a new store and making designs that loosely mimic the painting styles of famous artists, but (hopefully) cuter and fun rather than serious masterpieces. (Don't think I've got any masterpieces in me yet!)

The question is how to do this in a way that doesn't trip any filters on Zazzle. I'm guessing designs "in the style of" a particular artist are fine, as long as I do not mention that artist's name in any of the metadata (title, description or tags)?

I managed to get a bunch of my designs deleted by Zazzle a long time ago, as they kind of had an Andy Warhol vibe to them, so I said so in the description.

Just wondering what other Zazzlers - or Zazzle Staff if they have time to chip in - think about this in a more general context. I'm guessing it has to be considered on an individual basis, maybe?

Warhol, as a modern artist, probably has an estate with an army of lawyers protecting his name; whereas Da Vinci might be, dare I say it, mentionable in the metadata. I'm guessing Dali would be similar to Warhol. Maybe I should just use the style descriptor as a proxy (something like "pop art", "surrealist", "renaissance", etc)

Any thoughts (before I dive headlong into a new project, only to get it busted for a dodgy tag or description saying something I'm not allowed to say)?

Thanks in advance.

[KeegansCreation]

You're right about Dali. His estate will get your design yanked if you say "in the style of Dali" in the description, title or tags. As with Warhol, it isn't the design itself, it's the name that gets the design yanked. Although if you imitate Warhol by painting a recognizable commercial product like Cambell's soup or person like Elvis, that could get it yanked too. If you can pull off painting hyper realistic yet surreal paintings as Dali did then go for it, but don't mention his name.

Style descriptors such as "pop art", "surreal" and "renaissance" are fine in the titles, tags and description.

Artists of the 19th century or earlier can be noted by name. I have designs that I loosely adapted from William Morris textiles and say so in the description, title and tags. But he's 19th century.

If you are going to use a photo as a reference for your painting, make sure that photo is in the public domain. Warhol's estate was sued by a photographer for using her copyrighted photo of Prince as a reference and Shepard Fairey was also sued for his use of a copyrighted photo of Obama as a reference.

[Artsiren]

Thanks KC! I feared that - but, hey, they created their fame and popularity. I can understand them (or their descendants) not wanting others to cash in on the family name.

It's all going to be very experimental (like most stores on my Artsiren account) - using AI generation, but with multiple generators, and alongside my own modifications. Mainly it will start with Text-to-Image AI, and then that will serve as the "reference", so I should be okay with that; if I need to add any new elements, I'll take your advice and only use public domain images for them.

Thanks again!!

[SimplyDesigned]

You need to make sure these ai text websites you plan on using actually allow you to use the photos produced commercially. It's a grey area in most places you really need to read the fine print to make sure you're good to go.

[Artsiren]

Thanks SD. That's a really excellent point! I'll mainly be using Stable Diffusion and derivatives of it for now, so the rights to Input/Output are fine. I think the grey areas appear when people producing the artistic works want to assert copyright. As you imply, it'll be a matter of law where the person tries to do this, as it will usually require "human creative endeavor".

If the answer is yes it does, one can get the copyright; if no, it is Public Domain (for Stable Diffusion, at least). So I guess the latter case is the same as using CC0 fragments for any fully human created design. Although with CC0 images, I guess if you do enough work on it, it can be copyrightable in its own right (so, back to creative endeavor).

The thing I'm having to be careful about with Stable Diffusion is that it doesn't product any Output that resembles an image in the training datasets. There's a way to check that though, which I'll be using. But the eventual aim will be many layers and a lot of layer masking, both to add creative human endeavor, and also so that large portions appear directly from my own mind, rather than just a single sweep through a training dataset.

Not sure I can post external links here, but an artist called Albert Bozesan did a great example on his Youtube channel (Title of video: How to Make Concept Art with AI (Free and Easy) - Stable Diffusion Tutorial 2022). So it's not exactly a case of Press the GO Button, drink coffee, then Publish a Design. At least if you want nice results - his are pretty astonishing pieces of art. I hope I can get anywhere near his standard, with practice.

Thanks again!

[KeegansCreation]

This sounds like a great project. I have a friend who is very into Stable Diffusion and Dall E and regularly posts images made by himself and others on facebook. He doesn't sell them, just posts them, so the issue of copyright for selling doesn't come up. One image was in the style of a medieval manuscript except with Cthulu in place of Jesus. AI seems adept at "in the style of" images. The developers probably use a lot of classic art pieces for training it.

Doing radical transformation with masked layers sounds like a good way to go. I can see making several different AI images and then using masking to bring them together into a new image- one that wouldn't be CC0.