Stable Diffusion 3.5
Stable Diffusion 3.5 has impressive prompt understanding and adherence skills with its high-quality image creations. This guide will provide a detailed explanation of how to use SD3.5.
Last updated
Stable Diffusion 3.5 has impressive prompt understanding and adherence skills with its high-quality image creations. This guide will provide a detailed explanation of how to use SD3.5.
Last updated
Stable Diffusion 3.5 is a new base model created by Stability AI with 8.1 billion parameters. The model has impressive prompt understanding and adherence skills, producing high-quality image creations. SD3.5 has great knowledge of mediums, styles, artists, mangakas, illustrators, and more, making it very flexible in its usage.
Try the Model: Stable Diffusion 3.5
New Stable Diffusion has some similarities to Flux model in prompting, this means SD3.5 is also tend to natural language prompting and doesn't allow some features as Flux did before
Keyword Weighting does not work. e.g. (term:1.3), [[term]], (((term)))
Keyword Blending/Edit tools does not work.
BREAK does not effect.
Better understanding of natural language prompts.
It can create different subjects in image without any effort.
e.g. dog and cat generates two different anatomically correct animal on Stable Diffusion 3.5 while most of other models tends to generate creatures that are a combination of these two animals. However, Stable Diffusion 3.5 doesn't ignore negative prompts. Negative prompts are not necessary for model to work good but users still have option to use unlike the Flux. Also it has outstanding art, artist style knowledge compared to Flux, as many of you know Flux was pretty weak at understanding references and recognizing them but SD3.5 still have strong database of styles like its older versions. This makes SD3.5 great option for users couldn't find enough artist knowledge and beauty in Flux.
Generating high quality images without any "Quality Modifier" prompt.
Generating desired images with correct details and placements even with confusing prompts cause problem on other models. e.g. blue cube under red ball
Generating accurate text renders in images. It can struggle with very long texts but capability of model is enough to satisfy most of users.
Generating almost any subject with correct prompting.
Crafting images with any art style or medium, it has great dataset of art styles.
Crafting ultra photorealistic images that are indistinguishable from real photographs but Flux is better than SD3.5 in a terms of realism.
Generating correct anatomy and hand renders but it still struggle with certain parts of body such as neck, chin, cheekbones etc.
Crafting images with SFW content.
Recognizes Art/Artist styles from references.
Works great with both of natural language and tag based prompts. Long descriptive natural language prompts are recommended for detailed compositions.
Don't use brackets in your prompt.
You can use it from AI App or standart generation page.
Works good with base SDXL resolutions.
It's not best model to generate NSFW, better to wait for LoRA support or new SD3.5 based models.
Text must be used inside of quotation marks ".
Negative prompts are not necessary but still can be used for unwanted stuff.
Steps: 20 (can be increased up to 40)
CFG Scale: 4.5 (7 can be used for anime, cartoon like arts)
Sampling Method: euler+beta, euler+sgm_uniform, dpmpp_2s_ancestral+sgm_uniform, dpmpp_2m+sgm_uniform
SD 3.5 has strong natural language prompt understanding capabilities with its ease of use style, model also supports negative prompt box so we can edit our images in easier and more effective way compared to Flux. It works good with tag like prompting but natural language prompts are recommended for best outputs.
Stable Diffusion 3.5 has outstanding text renders compared to other AI image generators, user must write text inside of quotation marks " to seperate it from prompt. e.g.
text written on a road sign "SeaArt"
Style/Medium
Subject
Subject Details
Art/Artist References
Additional Details
Lighting/Shadow
Additional Modifiers
Start with your subject and describe subject's details, style and medium. Then add additional details to create composition of image such as lighting, background, effects. After that you can add last modifiers to complete your prompt.
Note: Dot . can be used inside of prompt after finishing sentences.
Use natural language prompt and don't seperate every word with comma , then reinforce composition of image with additional tags.
Positive Prompt
A detailed portrait of a woman with flowing hair and floral motifs, in the style of Alphonse Mucha, art nouveau, intricate patterns, iridescent watercolor ink splashes, high contrast colorful artwork, brush lines over image that remind us confusion of life.
Negative Prompt
nsfw, nudity, child, childish
Positive Prompt
Cinematic photography, A cozy winter cabin nestled in a snowy forest, soft golden light glowing from the windows, surrounded by tall pine trees. Realistic style, high detail, warm atmosphere, winter wonderland scene.
Negative Prompt
blurry, dark, empty.
Positive Prompt
A dark, ominous medieval landscape with a lone warrior holding a large sword, in the style of Kentaro Miura, highly detailed, "Embrace The Darkness" written in handwrite on the sky, intense shadows, dark fantasy.
Negative Prompt
nsfw
