Hello,
I have spent a lot of time trying to create a system prompt that will enable Pickaxe to generate images in different formats, particularly 16:9. Unfortunately, I have not been successful so far. I’ve started to think that the issue might be with Pickaxe itself, as none of the numerous prompts I’ve tried have worked. Could we have an option to give Pickaxe the ability to generate images in formats other than the current square format?
Thank you.
Hey Kent, you are correct. Right now Pickaxe only supports 1x1 image generation. We may expand our image generation capabilities further in the future as right now we’re mainly focused on language models.
I’ve thought ChatGPT 4 and 4o can render images in 16:9 and 9:16 all you have to do is append the prompts with (1792x1024) or (1024x1792) at the end.
What AI is rendering the images in pickaxe (with Flux not selected)?
Flux is the default (free) image model. You can use DALL-E if you connect it as an Action. Currently both of those models only output images with a 1x1 aspect ratio on Pickaxe. That’s due to our configuration.
If you connect to MidJourney you can generate in a variety of aspect ratios.
Is the 1:1 ratio a part of how Flux works? I’m not sure what’s involved in supporting dynamic aspect ratios but it seems like the square ratio limitation is significant when so much media is closer to a 9:16 or 16:9 ratio.