Official DALL-E Gallery: October 2024 (Halloween Edition)

A nice toon vampire :slight_smile:

9 Likes

I’m sorry guys, I didn’t start here, just rolled some ideas together

Pushing the limits a bit

4 Likes

Super cute! Her demeanor is excellent, and her face speaks volumes.:rabbit::honeybee::heart:

1 Like

You inspired me.

wow! Who would of thought 🧐

5 Likes
They.. are... EVERYWHERE...!!!

4 Likes

Got my Halloween decorations up in the yard…

A whimsical playhouse inspired by the folklore of Baba Yaga, perched on stilts resembling chicken legs. The house is small, with pastel colors including light pink and pale blue, and quirky windows and doors. There is a sign hanging on the front door that says ‘Beware the Baba Yaga!’ The legs are wooden, shaped like chicken feet, lifting the house off the ground. Two black birds are sitting on the roof. The setting is a backyard with green grass and fall leaves scattered around.

7 Likes

Love it her hut is one of my fav AD&D artifacts I ran a whole adventure in it. Early TARDIS tech :rabbit::jack_o_lantern:

who?

An owl in the dark its eyes aglow, a haunted twisted gritty dark wood. Depth of perspective centered as if a tunnel

4 Likes
Halloween creatures having Halloween... simply tooo much!!!

4 Likes

Wonderfully done :heart::jack_o_lantern::ghost::skull: it is like the bright excitement of imagination the day before the night of trick or treating.

But as night falls…

A night party of skeletons celebrating the darkness Halloween offers night dance Mid range view party once bright decorations now rotten and decrepit as Halloween night brings fright wide image

4 Likes

Nice. Have you found using JSON prompts to be more effective (consistent) than plain text prompts?

3 Likes

I used your prompt with a little modification. It worked!
I never tried to structure prompts like JSONs but could it be it is more precise, or is DallE just reading all the words out?

@mitchell_d00

Catch me if you can

4 Likes
Happy Halloween!🎃

If I tinker with it I can play with most images just by breaking down image logic.
“ Old Wood table a red and green sweater arm wearing a leather work glove, long sharp single edged blades mechanically attached at the tips of 4 fingers and the thumb of the glove like claws spread out over a page of antique parchment on it exactly “happy Halloween Daller” wide image”


Old Wood table a red and green sweater arm wearing a leather work glove, long sharp single edged blades mechanically attached at the tips of 4 fingers and the thumb of the glove like claws spread out over a page of antique parchment on it exactly “happy Halloween Daller” close up of single hand wide image

2 Likes
i added more artistic perspective by adding a hat I love image logic 🎃❤️

Old Wood table on it a black fedora , a red and green sweater arm wearing a leather work glove, long sharp single edged blades mechanically attached at the tips of 4 fingers and the thumb of the glove like claws spread out over a page of antique parchment on it exactly “happy Halloween Daller” close up of single hand wide image


2 Likes

Very cool, tank! :jack_o_lantern:

2 Likes

Yes, generally speaking I have found that reducing a prompt to its core elements, converting to json and adding in the desired parameters assists in more consistent generation.

Testing it with your prompts, and continually reducing the parameters until the generation degrades, is also a good way to determine what is and is not superfluous in your prompt.

To keep things on topic, if you and @Daller are interested, I can start a thread specific to using json for Dall-E and we can get into the weeds over there.

4 Likes

If you make it I’d be happy to join in. It sounds like a lovely idea. Image structure intrigues me. We have a lot of threads about the data of images but not many about the underlying frames of structure.

2 Likes

Yes, would be cool, i will link it on the thread with the tips.

5 Likes
3 Likes

I’d also be interested in a separate topic on this.

3 Likes