Thank you for sharing. I have two questions. Will the result of the generated images always be a square with 4 images? Is it possible to apply the sref codes to an existing image instead of generating a result from a code?
Thanks for watching and great question. I'm not a lawyer so I can't really comment on the specifics of the legality. AI copyright has a lot of gray areas and you'd be better off asking a professional attorney that question. I honestly don't know.
Thank you so much for this. I noticed that when I use shared Sref codes on discord rather than MJ site, I get better results? At Mj site it often ignores the Sref and creates a random style, usually close to realistic. Do you know why?
I’m new to AI and have a question. If I create an illustration with these techniques you demonstrated can I legally use them on a printed brochure for my business?
Thanks for watching and great question. Disclaimer: I'm not a lawyer. What I say or type should never be intended as legal advice. I recommend reaching out to a copyright attorney for proper legal advice. End of disclaimer. Also, check out this article. It does a great job at explaining the possibilities of legal claims with AI generated artwork. I recommend reading the "How to Avoid Infringement" section: copyrightlately.com/using-ai-artwork-to-avoid-copyright-infringement/ With that being said, there is a lot of grey area on this topic. However, when referencing images, the generated output could still infringe one of the copyrighted elements when found within a preexisting work. Additionally, prompts that refer to the style of a particular living artist could result in a legal claim, depending upon how similar the work is to one of that artist’s originals. Hope that helps!
Hi Ben, thanks for your video. Is it possible to copy a style with --sref to a photo we already have instead of having to put a text promt over an idea or does it only allow it if we put text in the prompt?
Yes, the --sref code in MidJourney can reference a style from an existing image. You’ll typically need to provide a text prompt to clarify your desired output, but the image reference (via --sref) plays a significant role in influencing the result. Unfortunately, entirely skipping the text prompt isn’t supported at this time.
Thanks for watching! From my understanding, every possible --sref code corresponds to an existing style, but you’re not creating the style. Instead, each code points to a specific position within the AI model's latent space (essentially the “map” of all images it can produce). The style was already present in the model; you simply retrieved it. When people share --sref codes, they’re not presenting styles they invented; they're showcasing styles they discovered, either by using --sref random or by entering a random string of numbers. I hope that answers your question and thanks for watching!
@@blogwithbenHe’s talking about the code of those --sref (random number) and how he can create his own. In your video, it shows how to make a --sref prompt based on an image, but not the code for it to use for future prompts.
Hello, thank you for the video, but I was expecting a more detailed explanation. I was hoping for a guide that used different combinations, such as using multiple reference images, using different weights for reference images and using both reference images and the random parameter. Thank you anyway.
Thank you for sharing. I have two questions.
Will the result of the generated images always be a square with 4 images?
Is it possible to apply the sref codes to an existing image instead of generating a result from a code?
great vedio! I have a question, can I use all sref codes commercially?
Thanks for watching and great question. I'm not a lawyer so I can't really comment on the specifics of the legality. AI copyright has a lot of gray areas and you'd be better off asking a professional attorney that question. I honestly don't know.
Thank you so much for this. I noticed that when I use shared Sref codes on discord rather than MJ site, I get better results? At Mj site it often ignores the Sref and creates a random style, usually close to realistic. Do you know why?
I’m new to AI and have a question. If I create an illustration with these techniques you demonstrated can I legally use them on a printed brochure for my business?
Thanks for watching and great question. Disclaimer: I'm not a lawyer. What I say or type should never be intended as legal advice. I recommend reaching out to a copyright attorney for proper legal advice. End of disclaimer.
Also, check out this article. It does a great job at explaining the possibilities of legal claims with AI generated artwork. I recommend reading the "How to Avoid Infringement" section: copyrightlately.com/using-ai-artwork-to-avoid-copyright-infringement/
With that being said, there is a lot of grey area on this topic. However, when referencing images, the generated output could still infringe one of the copyrighted elements when found within a preexisting work. Additionally, prompts that refer to the style of a particular living artist could result in a legal claim, depending upon how similar the work is to one of that artist’s originals. Hope that helps!
@@blogwithben thanks!!
Hi Ben, thanks for your video. Is it possible to copy a style with --sref to a photo we already have instead of having to put a text promt over an idea or does it only allow it if we put text in the prompt?
Yes, the --sref code in MidJourney can reference a style from an existing image. You’ll typically need to provide a text prompt to clarify your desired output, but the image reference (via --sref) plays a significant role in influencing the result. Unfortunately, entirely skipping the text prompt isn’t supported at this time.
how can i create the sref code?
Thanks for watching! From my understanding, every possible --sref code corresponds to an existing style, but you’re not creating the style. Instead, each code points to a specific position within the AI model's latent space (essentially the “map” of all images it can produce). The style was already present in the model; you simply retrieved it.
When people share --sref codes, they’re not presenting styles they invented; they're showcasing styles they discovered, either by using --sref random or by entering a random string of numbers. I hope that answers your question and thanks for watching!
how do I create a unique numer? I see people sharing unique sref numbers and I only have the URL how can I do that? thanks
Thanks for watching! Where are you seeing that?
@@blogwithbenHe’s talking about the code of those --sref (random number) and how he can create his own. In your video, it shows how to make a --sref prompt based on an image, but not the code for it to use for future prompts.
Hello, thank you for the video, but I was expecting a more detailed explanation. I was hoping for a guide that used different combinations, such as using multiple reference images, using different weights for reference images and using both reference images and the random parameter. Thank you anyway.
Hey! Thanks for watching and for the feedback. I work on a more detailed video in the future. :)
@@blogwithben good luck bro