# stablediffusion nftart aiart omen aiia psychedelicart ai crackingthecode

test5 has slightly more training images than 1.3.0, and the output is close to 1.3.0, but degraded.
In 1.3.0, I did a 512x512 cropping step, but in test5 I did not do that, but used aspect ratio bucketing.
Maybe the same procedure with the exact same dataset would reproduce 1.3.0

0 1
2023-01-12

Yes...
I am currently testing a WD1.4 based model that dataset with different contents to verify why v1.3.0's outputs are so nicely.
Even if I increase the number of training images, I cannot exceed the miraculously created 1.3.0. I cannot reproduce it... 😭

0 1
2023-01-12

After weeks of trial and error to finetune my style using various techniques, datasets, and prompts, trying dozens of datasets / prompts, hundreds of settings and generated thousands of samples, I'm pleased with the results!

4 41
2023-01-03

Happy New Year one and all...and we are off to a roaring start for 2023. Globally, at +0.7°C above the 1979-2000 baseline, this is the highest anomaly I have witnessed for this dataset.

236 517
2023-01-01

We're surprised by how well it performs, and interestingly - the order in which the training data was organised had a significant impact on the final model.

Another training run on the same dataset in a different order didn't garner anywhere near the same consistency of results.

2 12
2022-12-22

I learned something interesting about A.I. art. The datasets used in Midjourney and Stable Diffusion were created by LAION, a non-profit ai research organization. (1/7)

0 2
2022-12-21

Let's try a prompt that's nowhere close to the dataset, how about "a zombie". 🧟‍♂️

This is interesting - the lower strength training (left) gives you a different feel with the pixel art and can deviate further stylistically, while the stronger runs (right) look more polished.

0 7
2022-12-20

Very soon AI art generators won't be inspired by any art (already less than 2% of the dataset) + it will still generate art with ease + the whole "stolen remixer" narrative will collapse.

Also AI will have no "tells" like crummy hands + it'll be perfectly coherent.

Then what?

48 483
2022-12-18

From my datasets
Flowers will grow
And this is AGI
And this is singularity

45 296
2022-12-18

A few from my latest collection. Trained from a dataset including my handmade wool felt artworks, paintings and nature photography.

1 6
2022-12-15

I was pretty shocked that when I put in the name of one of my favourite singers for a music video MidJourney actually generated some accurate pictures of her. I thought this would for sure be blocked or like, not be allowed to include her face in the dataset

0 0
2022-12-13

GM!

Still working on captioning this dataset. Hope your day is pleasant and productive!

5 26
2022-12-11

Yes, Twitter, I too agree that the inclusion of art in an AI dataset and claiming it doesn't exist because its 'data' Is pretty fucking ridiculous and deserves a warning, but not like this.

0 9
2022-12-07

Some recent artworks trained from my wool textile artworks datasets.

0 5
2022-12-07

It would be a shame if this effect were accelerated by the deliberate feeding of image/text errors into these datasets. Potato hands landscape concept ArtStation realistic big boob.

0 8
2022-12-07

本日はキャラクターアニメーションと画像生成系の採択論文セッションをしっかり聴講して、関連論文やリポジトリも掘っていたのですが、収穫だったのはこちら
”The DanbooRegion 2020 Dataset”
https://t.co/KFyFYb4fpr
アノテーション領域が与えられたDanbooru、といって伝わるだろうか。12人で評価。

8 20
2022-12-06

Close-up views

"Dataset engineering" >>> "prompt engineering"

4 52
2022-12-01

Is this Mofo serious? I already found several of my older artworks in LAION-5B dataset and I haven't even scratched the majority of it. This lying s+++ of sh+t

2 14
2022-12-01

Exact same prompt. *Almost* the same training dataset.

To the right, I added a few more pictures to add a bit of variability. The results are much better! (details, colors).

Prediction: "dataset engineering" will be more critical than "prompt engineering".

32 294
2022-11-21

We've just published a whole load of Land Cover Map 1km summaries for both GB and Northern Ireland.

The datasets are now available for 2017, 2018, 2019, 2020 and 2021

https://t.co/XQEAnVY6MG

43 152
2022-11-18