For image Consistency add the following after the instruction. “Always display GEN_ID.” The GEN_ID is a unique identifier for each image generated by DALL·E. It's useful for several reasons: Reference: If you want to discuss a specific image, mention improvements, or request variations, you can refer to it using its GEN_ID. This helps me understand exactly which image you're talking about. Tracking: It helps keep track of which images have been generated in a conversation, ensuring better organization and referencing. Revisiting: If you return to this conversation later and want to revisit a specific image, the GEN_ID makes it easy to identify. It's a handy tool for efficient and clear communication about the generated images!
Hi, Dalle-e 3 supports only 3. You cannot specify a resolution. You can use third part software to upscale. You can say tall, wide or square in the prompt and it will generate the following resolutions: 1. 1024x1024 (square) 2. 1792x1024 (wide) 3. 1024x1792 (tall)
You cannot specify a resolution. You can use third part software to upscale. You can say tall, wide or square in the prompt and it will generate the following resolutions: 1. 1024x1024 (square) 2. 1792x1024 (wide) 3. 1024x1792 (tall)
Let me clarify, the images with minor changes also won't be 100% the same. They will maintain 80-90% or higher consistency, like the shirt example I gave in the video. However, like the logo example asking for any change re-generates the image. It is still far better than MidJourney in terms of consistency because, unlike Stable Diffusion (100% replicable), whenever MidJourney Rolls out a cluster update, the --seed won't be the same image. This is because the cluster update means adding or removing processors/GPU cores that won't be able to generate the image (100% consistency) with --seed.
For image Consistency add the following after the instruction. “Always display GEN_ID.”
The GEN_ID is a unique identifier for each image generated by DALL·E. It's useful for several reasons:
Reference: If you want to discuss a specific image, mention improvements, or request variations, you can refer to it using its GEN_ID. This helps me understand exactly which image you're talking about.
Tracking: It helps keep track of which images have been generated in a conversation, ensuring better organization and referencing.
Revisiting: If you return to this conversation later and want to revisit a specific image, the GEN_ID makes it easy to identify.
It's a handy tool for efficient and clear communication about the generated images!
Very helpful video. My first attempts were less than impressive. (User faults, not Dall-E 3) Using your tips and tutorials helps immensely. Thank you.
Glad it helped!
cant you specify the tall resolution pre-generation???
Hi, Dalle-e 3 supports only 3. You cannot specify a resolution. You can use third part software to upscale. You can say tall, wide or square in the prompt and it will generate the following resolutions:
1. 1024x1024 (square)
2. 1792x1024 (wide)
3. 1024x1792 (tall)
how do you get specific resolution? say 512x768? thanks
You cannot specify a resolution. You can use third part software to upscale. You can say tall, wide or square in the prompt and it will generate the following resolutions:
1. 1024x1024 (square)
2. 1792x1024 (wide)
3. 1024x1792 (tall)
If i purchasing gpt plus, did i get dall e 3 for free? Or need token to generate?
It’s included with gpt plus. I did not purchase any extra token.
@@controlaltai did have limit to generate dall e 3? Like how much prompt/hr?
@zanmf I believe the limit is 50 messages every 3 hours.
Strange, whenever I ask corrections of an image it gives me a totally new image EVERY time, don't know why it listens to you lol
Let me clarify, the images with minor changes also won't be 100% the same. They will maintain 80-90% or higher consistency, like the shirt example I gave in the video. However, like the logo example asking for any change re-generates the image. It is still far better than MidJourney in terms of consistency because, unlike Stable Diffusion (100% replicable), whenever MidJourney Rolls out a cluster update, the --seed won't be the same image. This is because the cluster update means adding or removing processors/GPU cores that won't be able to generate the image (100% consistency) with --seed.