Download
1 variant available
7400 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
4520 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
2.2K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
(89)
Jan 31, 2025
Test version. Just wanted to see how big batch size and how many more steps can i train without getting unusable results for a relatively small dataset.
Anyway the style is better on this one than the previous version .
Trained for 64k steps but final epoch is not usable so uploading the 60k step model. (batch size 8)
(Multiple people will probably struggle with anatomy )
Show more

2080 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
9590 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
V3 is trained for 60k steps. Trigger word- 90LOG-PHOTOGRAPHY
V2 is trained for 40k steps. It only needs the trigger word - [90LOG-PHOTOGRAPHY] .
I tried creating a LoRA using a collection of 1990s images that I found on the internet. My goal was to capture the photography style of that era, but instead, it focused primarily on facial styles. The resulting look reflects the Asian aesthetic of the 1990s, rather than the modern doll-like, cute appearance seen today
I think the trigger 90log for the look, and low quality, old photo , grainy ,noise, uneven lighting, color fading , worn photo for the style gives good results.
