👋 If you like what I do and want to support the development, feel free to buy me a coffee:
I'm back, and this time I'm bringing my most advanced, flexible, and powerful system yet!
Since my last milestone release back on June 9, 2026 (based on Better Days v0.2), I’ve been quietly grinding behind the scenes. Driven by personal R&D and a few great discussions in DMs regarding Anima compatibility, I decided to test its limits and explored two distinct development paths:
• Extracting Anima and compiling it to be 100% compatible across SDXL.
• Engineering a hyper-optimized version built exclusively for the Anima environment.
Because these Anima-specific builds still require further testing and benchmarking before a public release, I’m keeping them in the lab a bit longer. For now, I decided to focus my full attention on perfecting and releasing the core modular system and files currently in my hands—ensuring you get maximum stability right out of the box.
While those previous builds did a fantastic job enhancing Velvette and other similar checkpoints, I felt something was still missing: that "recognizable spark"—the raw, unadulterated essence of the base render style.
However, locking down that exact style was no easy task—a single minor shift in tensor weights and the whole thing instantly becomes incompatible. That was the real battle, of course...
As a huge bonus discovered during these hard-fought experiments, this extraction framework opens a goldmine for nostalgia lovers: bringing back classic SD 1.5 styles into the modern era.
SD 1.5 holds years of legendary, community-crafted checkpoint aesthetics that were unfortunately lost or washed out during the industry transition to SDXL. With this new modular extraction methodology, we can now:
• Mine vintage, handcrafted render signatures and unique brushstrokes.
• Isolate pure visual DNA without carrying over old 512x512 resolution limitations or outdated anatomical flaws.
• Port them directly onto high-resolu