--- license: apache-2.0 language: - en tags: - SD3.5-medium - diffusers - lora - replicate - image-generation - StableDiffusion3.5 - StableDiffusion3.5-medium - photo - Mayakovsky - art - poster - realism - character - historical person - poetry - literature - history - archival base_model: "stabilityai/stable-diffusion-3.5-medium" pipeline_tag: text-to-image library_name: diffusers emoji: 🔜 instance_prompt: MAYAK style Constructivist Poster art widget: - text: Step 1400 | MAYAK style drawing output: url: 1730619601173__000001400_1.jpg - text: Step 3000 | MAYAK style output: url: 1730622478102__000003000_3.jpg - text: Step 3400 | MAYAK style drawing output: url: 1730623154193__000003400_0.jpg - text: Step 3200 | MAYAK style drawing output: url: 1730622834121__000003200_3.jpg - text: Step 4400 | MAYAK style drawing output: url: 1730624957944__000004400_2.jpg - text: >- St5k| MAYAK style art of poet Mandelstam reading /YOU'VE RETURNED HERE, SO SWALLOW THEN, FAST AS YOU MIGHT, ALL THE FISH OIL OF LENINGRAD'S RIVERINE LIGHT!/ output: url: 1730626012058__000005000_1.jpg --- # Mayakovsky Style Soviet Constructivist Posters & Cartoons Flux LoRA – Version SD3.5-i – by SOON® Trained via Ostris' [ai-toolkit](https://github.com/ostris/ai-toolkit) on 50 high-resolution scans of 1910s/1920s posters & artworks by the great Soviet **poet, artist, & Marxist activist Vladimir Mayakovsky**.
For this training experiment, we first spent many days rigorously translating the textual elements (slogans, captions, titles, inset poems, speech fragments, etc), with form/signification/rhymes intact, throughout every image subsequently used for training.
These translated textographic elements were, furthermore, re-placed by us into their original visual contexts, using fonts matched up to the sources.
We then manually composed highly detailed paragraph-long captions, wherein we detailed both the graphic and the textual content of each piece, its layout, as well as the most intuitive/intended apprehension of each composition.
This repo contains various step checkpoints and sampless. Training went for 5000 steps at a DiT Learning Rate of .0002, batch 1, with the Adamw8bit optimizer, and the text encoders trained alongside the DiT!
No synthetic data was used for the training, nor any auto-generated captions! Everything was manually and attentively pre-curated with a deep respect for the sources used.
This is a **rank-16/alpha-16 Constructivist Art & Soviet Satirical Cartoon LoRA for Stable Diffusion 3.5 Medium**
Check out our **translations of Mayakovsky's verse-works**, adapted from a proto-Soviet song-tongue into a Worldish one...
And found, along with many other poets' songs and tomes... Over **at [SilverAgePoets.com](https://www.silveragepoets.com)!** ## Trigger words You should use `MAYAK style` or `MAYAK style Constructivist poster art` or `MAYAK style satyrical drawing` or suchlike to summon the poet's deathlessly pens and brushes. For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co./docs/diffusers/main/en/using-diffusers/loading_adapters)