1. DALL-E has no negative prompts.
To remove an object (like a gun) from an image:
Use a positive prompt that is the opposite of the unwanted object. Example:
Doesn't work: "No gun"
Works: "Family friendly"
To remove an object (like a gun) from an image:
Use a positive prompt that is the opposite of the unwanted object. Example:
Doesn't work: "No gun"
Works: "Family friendly"
2. DALL-E inside ChatGPT can only generate one image at a time.
But if you want more images without wasting your chats:
Ask it to generate X images, *one after the other*.
Works like magic.
But if you want more images without wasting your chats:
Ask it to generate X images, *one after the other*.
Works like magic.
3. You can change the aspect ratio of the images created inside ChatGPT.
- Square: 1024x1024
- Wide: 1792x1024
- Tall: 1024x1792
This is not possible in Bing Image Creator.
- Square: 1024x1024
- Wide: 1792x1024
- Tall: 1024x1792
This is not possible in Bing Image Creator.
4. When writing prompts, be specific rather than vague.
Vague prompts confuse DALL-E and can often result in weird text appearing.
Bad: A wizard with the ability to cast fireballs
Good: A wizard holding a fireball in his hand
Vague prompts confuse DALL-E and can often result in weird text appearing.
Bad: A wizard with the ability to cast fireballs
Good: A wizard holding a fireball in his hand
5. When writing DALL-E prompts, do not put verbs like "imagine" or "create" at the start of the prompt.
ChatGPT likes to do this - I have found it actually confuses DALL-E.
No: Create a pencil sketch of a boy
Yes: A pencil sketch of a boy
ChatGPT likes to do this - I have found it actually confuses DALL-E.
No: Create a pencil sketch of a boy
Yes: A pencil sketch of a boy
6. DALL-E has no image prompts.
But you can upload an image to ChatGPT.
Then ask it to recognize the image, and generate one that is as similar as possible.
Kinda works.
But you can upload an image to ChatGPT.
Then ask it to recognize the image, and generate one that is as similar as possible.
Kinda works.
7. DALL-E likes to change your prompt and make it longer - this is true even for the API.
To make sure it uses the prompt you provide, add this at the start of your prompt:
"Use this prompt EXACTLY. DO NOT change or add anything."
Or use Bing Image Creator.
To make sure it uses the prompt you provide, add this at the start of your prompt:
"Use this prompt EXACTLY. DO NOT change or add anything."
Or use Bing Image Creator.
8. Some influential tokens that can greatly affect the output image:
- Artistic
- Intricate
- Minimalistic
- Vibrant
- Artistic
- Intricate
- Minimalistic
- Vibrant
9. You can split an image into multiple images inside the prompt.
3 example prompts:
- 3 different comic panels featuring a knight in a cave
- character sheet featuring a wizard from multiple angles
- 4 different images showing the life cycle of a rose
3 example prompts:
- 3 different comic panels featuring a knight in a cave
- character sheet featuring a wizard from multiple angles
- 4 different images showing the life cycle of a rose
10. Words at the front of the prompt are emphasized more than those at the back.
"A vibrant red and yellow logo featuring a tiger"
Will be redder than
"A vibrant yellow and red logo featuring a tiger"
"A vibrant red and yellow logo featuring a tiger"
Will be redder than
"A vibrant yellow and red logo featuring a tiger"
11. Bonus:
To improve the accuracy of text generation, place the text to be generated at the START, not the end of the prompt.
To improve the accuracy of text generation, place the text to be generated at the START, not the end of the prompt.
That's all for now!
Hopefully you found this useful.
If you'd like to see more high quality content like this, consider following me at @chaseleantj.
Hopefully you found this useful.
If you'd like to see more high quality content like this, consider following me at @chaseleantj.
And check out my Logo Creator GPT if you haven't already :)
chat.openai.com
chat.openai.com
جاري تحميل الاقتراحات...