Stable Diffusion
Ranked 9 of 22 image generation models by recommenders
Free, open image generator people run locally and tune for fantasy and art; critics say its early outputs were weak and its safety filtering limited results.
Free, open image generator people run locally and tune for fantasy and art; critics say its early outputs were weak and its safety filtering limited results.
44 people gave an opinion
- recommend it
- 27
- criticize it
- 17
- approval share
- 61%
Recommended in 12 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 27 people recommend itNeeds 10 or more, met.
- 1.5 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 12 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Runs locally on consumer GPUs
“In many cases I actively prefer running Stable Diffusion locally because that's easier to fix the last 5%”
Comment from 26 Jun 2026 in “The AI backlash is only getting started” “it's like a production system. We can continue to tune and squeeze 1mp pictures out of it in high quality”
Comment from 12 Aug 2026 in “Automatic1111 for Apple metal, 40% speed up sd1.5” “It seemed so promising, when SD came out and it turned out the model fit on people's GPUs.”
Comment from 25 Mar 2026 in “OpenAI set to discontinue Sora video platform” - Comment from 16 Nov 2025 in “AI is killing privacy. We can't let that happen”
- Comment from 23 Dec 2025 in “Local AI is driving the biggest change in laptops in decades”
Strong results with fine-tuned checkpoints
“For art purposes SD1.5 with specialized and finely tuned checkpoints will still provide the best results by far.”
Comment from 26 Nov 2025 in “FLUX.2: Frontier Visual Intelligence” “it's like a production system. We can continue to tune and squeeze 1mp pictures out of it in high quality”
Comment from 12 Aug 2026 in “Automatic1111 for Apple metal, 40% speed up sd1.5” Fun, imaginative output for fantasy
“I never got it past outputting total nightmare fuel, but it was fun!”
Comment from 6 Jun 2026 in “What was your "Oh Shit" moment with GenAI?” “that thing was the bees knees for mi, imagination on a machine!”
Comment from 6 Jun 2026 in “What was your "Oh Shit" moment with GenAI?” “It's great for fantasy though, which is my main usage”
Comment from 2 Mar 2026 in “Microslop Manifesto” - Comment from 27 Feb 2026 in “Nano Banana 2: Google's latest AI image generation model”
Quick, controllable iteration and inpainting
“It's powerful enough that it encodes some really rich semantics, but small and light enough that iterative hacking on it is quick enough”
Comment from 26 Nov 2025 in “Image Diffusion Models Exhibit Emergent Temporal Propagation in Videos” “In many cases I actively prefer running Stable Diffusion locally because that's easier to fix the last 5%”
Comment from 26 Jun 2026 in “The AI backlash is only getting started” “you can do much more controllable inpaints with the latter”
Comment from 14 Jan 2026 in “AI Generated Music Barred from Bandcamp”
What people warn about
Each mark is one comment
Censored, safety-filtered outputs
“Stable diffusion 3 was so safe you couldn't generate a woman laying down on some grass”
Comment from 22 Dec 2025 on Hacker News “was laughed at at release for not being able to even generate a woman laying in grass”
Comment from 1 Jul 2026 in “Claude Sonnet 5” “censoring nudity from a model both compromises the model's quality (example: SD3)”
Comment from 12 Jan 2026 in “Ireland fast tracks Bill to criminalise harmful voice or image misuse” Output often bad or dated
“I found Stable Diffusion 1.5's output quite _bad_, say, but not as, I dunno, objectionable and wrong-looking as current models.”
Comment from 11 Nov 2025 in “Using Generative AI in Content Production” “Early Stable Diffusion set a lot of weird expectations that just aren't the standard now.”
Comment from 3 Mar 2026 in “AI-generated art can't be copyrighted (Supreme Court declines review)” “There's not much of a reason to use SD 1.5 over SDXL if image quality is paramount.”
Comment from 26 Nov 2025 in “FLUX.2: Frontier Visual Intelligence” Weak at text and fine details
“Stable Diffusion is much easier to deal with in that regard, but then it can't do text”
Comment from 18 May 2026 in “Where Are the Vibecoded Photoshops?” “Always ended up with the beak coming out where the visor glass should've been”
Comment from 12 Feb 2026 in “GLM-5: From Vibe Coding to Agentic Engineering” Too hard for most people
“Stable Diffusion and local models are still way too hard for 99.99% of people”
Comment from 25 Nov 2025 in “FLUX.2: Frontier Visual Intelligence” “Just keeping rails in parallel is beyond many Stable Diffusion type systems.”
Comment from 27 Aug 2026 in “507 Mechanical Movements”
Other image generation models people recommend
The four most recommended other image generation models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| Nano Banana | 265 | 159 | 63% | 12 | 6 mo5 owners | — |
| gpt-image-1 | 79 | 79 | 50% | 12 | — | — |
| FLUX.2 | 66 | 51 | 56% | 10 | — | — |
| gpt-image-2 | 63 | 26 | 71% | 8 | — | Approved |
| Stable DiffusionThis page | 27 | 17 | 61% | 12 | — | — |
What people said
40 of the 50 comments that mention it, newest first. Each links to the original comment.
“the first passable Stable Diffusion images”
“immensely useful new tech and research that benefits many”
“it was just way too easy and didn't require any effort”
“it was going to change the way people design, visualize and present things forever”
“Just keeping rails in parallel is beyond many Stable Diffusion type systems.”
“peering at a Stable Diffusion/Midjourney/Whatever image on one's screen”
“it's like a production system. We can continue to tune and squeeze 1mp pictures out of it in high quality”
“was laughed at at release for not being able to even generate a woman laying in grass”
“In many cases I actively prefer running Stable Diffusion locally because that's easier to fix the last 5%”
“For some the aspect ratio was impossible to meet, and it would fail if the banner was 200x60.”
“We've been flooded with low artistic value output and people are having a natural reaction to that.”
“that thing was the bees knees for mi, imagination on a machine!”
“Especially the image to image, and the first gos people had at making videos with it.”
“I never got it past outputting total nightmare fuel, but it was fun!”
“Stable Diffusion is much easier to deal with in that regard, but then it can't do text”
“Definitely one of the coolest inventions to ever come along in my lifetime”
“It seemed so promising, when SD came out and it turned out the model fit on people's GPUs.”
“the initial base model wasn't Stable Diffusion and wasn't trained internet scrapes without the copyright permissions”
“Early Stable Diffusion set a lot of weird expectations that just aren't the standard now.”
“It's great for fantasy though, which is my main usage”
“it still can't get historical generation right”
“It doesn't take much effort to get some art with a sense of originality or flow from Stable Diffusion”
“Always ended up with the beak coming out where the visor glass should've been”
“Stable diffusion does this quite handily.”
“Older Stable Diffusion models were preferred over newer, neutered models”
“When Stable Diffusion isn't good enough, no amount of re-prompting gets you something it can't do.”
“you can do much more controllable inpaints with the latter”
“censoring nudity from a model both compromises the model's quality (example: SD3)”
“Suno is kinda bad at werid instructions (like Stable Diffusion for images)”
“You can easily run models like Mistral and Stable Diffusion”
“Stable diffusion 3 was so safe you couldn't generate a woman laying down on some grass”
“If I generate art using DallE or Stable Diffusion, then edit in Krita/Photoshop/etc. it doesn't suddenly cover up the fact that I was unable to draw/paint/photograph the initial concept.”
“ChatGPT will profusely apologize and provide extremely detailed instructions in setting up local Stable Diffusion as an alternative”
“It's powerful enough that it encodes some really rich semantics, but small and light enough that iterative hacking on it is quick enough”
“There's not much of a reason to use SD 1.5 over SDXL if image quality is paramount.”
“For art purposes SD1.5 with specialized and finely tuned checkpoints will still provide the best results by far.”
“Stable Diffusion and local models are still way too hard for 99.99% of people”
“Llama, Stable Diffusion, and Wan Video are a glimpse of what's to come with local AI.”
“I actually really like Stable Diffusion 1.5.”
“I found Stable Diffusion 1.5's output quite _bad_, say, but not as, I dunno, objectionable and wrong-looking as current models.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.