<h3 id="concept">Concept</h3>This is an Expression LoRA trying to mimic the appearance and expression of a girl you might run into after 6 a.m. at a party, rave, or club. It also works well for the aftermath of a sports match.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/20037fd8-b5af-41d1-9cdb-641bf73919c5/original=true/20037fd8-b5af-41d1-9cdb-641bf73919c5.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/3a580e6f-c300-4a4f-a1c5-46639d2f94bb/original=true/3a580e6f-c300-4a4f-a1c5-46639d2f94bb.jpeg" /><h3 id="usage">Usage</h3><ul><li>Trigger Word: hpg_cm</li><li>Recommend strength: around 0.8</li><li>You may enhance the effects with keywords such as exhausted, tired, or sweaty.</li><li>Works fine with other character LoRAs as far as I've tested.</li></ul><h3 id="krea2-no-makeup-version-added">Krea2 no-makeup version added</h3>This version is essentially identical to Krea2 v1.0, but it has been trained using a dataset that does not feature smeared makeup. Some setups just work better with no makeup.<h3 id="zit-v1.0-vs-zit-v2.0">Zit v1.0 vs Zit v2.0</h3>Zit v2.0 was trained using a new dataset as a character preset, with the special traits (e.g. tiredness, sweatiness and smeared makeup) omitted from the captions.In contrast, Zit v1.0 was trained using a dataset as a concept preset, with the special traits explicitly mentioned in the captions.Just sayin'... It's up to you to decide which one to take. For me v2.0 feels more natural, more out of the box, no need for extra trigger prompts, and it definitely needs less strength.<h3 id="notes">Notes</h3><ul><li>Trained on Civitai with Z-Image Turbo / Krea2 as base model.</li><li>Showcase images are PNGs from ComfyUI that typically include the workflow.</li></ul>Have fun.

HardPartyGirlz - CrashMode

v2.0 was trained using a new dataset, with a character preset and a different caption concept.

Files with SHA256: f965163645806ba4424c9d8341cba555cc063a990478eb1a0a1e53728d3f46ff