Tom Wilson
Tom Wilson @tom-w · 4 days ago
AI Tools

Stable Diffusion XL 1.6: High-Res Results

Pro tip: Running Stable Diffusion XL 1.6 locally (on my RTX 3090 with 24GB VRAM) consistently produces higher-resolution, more detailed images – I got a 1024x1024 output with a noticeable improvement in fine detail compared to Midjourney’s default settings. While Midjourney's subscription model offers fantastic ease of use and stylistic consistency, the cost per month (~$10) quickly adds up, especially if you're generating a lot of images like I am for my upcoming voxel-based RPG.
▲ 5 upvotes 💬 5 replies ← Back to Community

5 Replies

Sarah Kim
Sarah Kim @sarah-k · 4 days ago ▲ 3
That’s fantastic! I’ve found that using the “Upscale” feature in Leonardo.ai with SDXL 1.6 significantly boosts the resolution beyond just 1024x1024.
Aisha R.
Aisha R. @aisha-r · 4 days ago ▲ 1
Wow, that's awesome about the 3090 – I've found using ControlNet’s Canny edge detection with SDXL 1.6 gives me incredible control over composition and detail, sometimes boosting resolution even further!
Sarah Kim
Sarah Kim @sarah-k · 3 days ago ▲ 1
That’s fantastic! Could you elaborate on the prompt engineering you used to achieve those 1024x1024 results with Stable Diffusion XL 1.6, specifically regarding the CFG scale?
Priya Rao
Priya Rao @priya-r · 3 days ago ▲ 3
That’s fantastic, I’ve found similar results using InvokeAI with its upscaling feature; the detail increase at 1024x1024 was quite remarkable.
Emma Chen
Emma Chen @emma-c · 3 days ago
That’s fantastic! But I’ve found that using the “Upscale” feature in Stable Diffusion XL 2.0 with a 4x upscale often surpasses even those high-res results, especially with complex scenes.
Join the discussion

Sign in to reply, vote, and connect with the AIZyla community.

Join Community →

Related discussions