Show HN: PromptHero – Search millions of prompts for Stable Diffusion and DALL-E (opens in new tab)

(prompthero.com)

56 pointsrameerez3y ago24 comments

24 comments

Also see lexica.art

Its recently updated to use CLIP for image search. This means now you get images based on description, and also matching images, allowing you to find a prompt to generate certain kind of results.

You might want to do something similar here.

freediver3y ago

Another vote for lexica.art, extraordinary execution.

sam_goody3y ago

There are some great examples here of both how far Airt has come and how far it still has to go.

Eg: https://prompthero.com/prompt/42a1e1dad9f - The description begins: "photo of a gorgeous woman Extremely detailed..".

The images are of a nice (blurry) den, but there is no-one in it.

Axsuul3y ago

I’m not seeing that many prompts for generating logos or icons. Anyone have any recommendations for that?

rameerezOP3y ago

I'm not sure these models are good at generating great logos yet, but searching for "logo design" returns a few good ones for me https://prompthero.com/search?q=logo+design

motyar3y ago

Looks good! What is your tech stack?

rameerezOP3y ago

Thanks! It's Ruby on Rails with some vanilla JS here and there when I need it.

scifibestfi3y ago

Millions already! Where do all the prompts and images come from? How are they collected?

frozencell3y ago

Discord.

scifibestfi3y ago

For Midjourney certainly, but Stable Diffusion and DALL-E?

1 more reply

santa_boy3y ago

which discord?

1 more reply

Pr0ject2173y ago

Pagination is broken.

rameerezOP3y ago

Where?

djtidau3y ago

https://prompthero.com/category/photorealistic?page=2

As mentioned above, the results do not change when paginating through categories. Page 3 is the same asPage 2 is the same as Page 1 etc..

1 more reply

cinntaile3y ago

How do I filter the results on model type?

rameerezOP3y ago

Search anything – the model selection filter is at the top right

cinntaile3y ago

Oh ok, thanks. The Stable Diffusion results don't show for me, the other two do.

1 more reply

j / k navigate · click thread line to collapse

24 comments

smusamashah3y ago

Also see lexica.art

Its recently updated to use CLIP for image search. This means now you get images based on description, and also matching images, allowing you to find a prompt to generate certain kind of results.

You might want to do something similar here.

freediver3y ago

Another vote for lexica.art, extraordinary execution.

sam_goody3y ago

There are some great examples here of both how far Airt has come and how far it still has to go.

Eg: https://prompthero.com/prompt/42a1e1dad9f - The description begins: "photo of a gorgeous woman Extremely detailed..".

The images are of a nice (blurry) den, but there is no-one in it.

Axsuul3y ago

I’m not seeing that many prompts for generating logos or icons. Anyone have any recommendations for that?

rameerezOP3y ago

I'm not sure these models are good at generating great logos yet, but searching for "logo design" returns a few good ones for me https://prompthero.com/search?q=logo+design

motyar3y ago

Looks good! What is your tech stack?

rameerezOP3y ago

Thanks! It's Ruby on Rails with some vanilla JS here and there when I need it.

scifibestfi3y ago

Millions already! Where do all the prompts and images come from? How are they collected?

frozencell3y ago

Discord.

scifibestfi3y ago

For Midjourney certainly, but Stable Diffusion and DALL-E?

1 more reply

santa_boy3y ago