This is early experimental LoRa for researching borders of training the concept of strict mirror.I think it is possible because SDXL can draw mirror floors.<h2 id="status-3nkor43w2">Status</h2>IL V2: this LoRa version can increase probability (+24% for 8-20 steps) of getting strict reflections in a new generation. Increasing steps is refining the results. Increasing LoRa's weight can refine the results (60% success with weight=2.0).IL V1: this LoRa version can slightly increase probability (+10% for 4 dmd2-steps, +12-16% for 8 dmd2-steps) to find strict reflections in a new generation. Increasing steps is refining the results.I go to the next step of the experiment.<h2 id="changelog-qx4qz9fgh">Changelog</h2><h3 id="il-v2-imi4bk1a1">IL V2</h3><ul><li>Added negative tokens:<ul><li>x_out_of_mirror (protects against a reflection outside mirror's bounds)</li><li>x_differect_reflection (supress a wrong symmetry)</li><li>different reflection (same the x_out_of_mirror but danbooru tag)</li></ul></li><li>Added ai-generated images with negative tokens to dataset.</li><li>60 epochs instead of 10 epochs of version 1</li><li>Trained with cleaned dictionary: x_mirror_reflection, x_reflection_face, x_reflection_facing_aside, x_reflection_facing_away, reflection, mirror, hand on mirror, full-length mirror, broken mirror, reflection focus, looking at mirror.</li></ul><h2 id="usage-5k4r5sqmi">Usage</h2>Connect LoRa in your workflow and then activate by word x_mirror_reflection in your prompt. Next choose one of these controls for the position of the character's face relative to the mirror:<ul><li>x_reflection_facing_away</li><li>x_reflection_facing_aside</li><li>x_reflection_face</li></ul>Now you can use negative prompt (+8% to find good mirroring in IL V2): x_out_of_mirror, x_differect_reflection, different reflection

Strict mirroring

1st iteration.I used DMD2 and SPO LoRas as accelerators, TCD sampler, sgm uniform as scheduler, [holy mix] Illustrous model.It isn't strict enough but I hate to wait.Metrics: -------------------------------------4 steps, with lora, 100 runsprompt: -------------------------------------artist, location, character,masterpiece, best quality, amazing quality, very aesthetic, absurdres, newest,(safe rating),x_mirror_reflection, {x_reflection_face, {|touching mirror}|x_reflection_facing_aside|x_reflection_facing_away},mirror, reflection, {1girl|1girl|__character__, 2girls}, symmetrical reflection, pose is symmetrical for the reflection,{dynamic pose|lazy pose|sitting}-------------------------------------total: 14% - strict reflection9% - strict reflection but outside mirror borders51% - wrong symmetry26% - trash (prompt misunderstanding)------------------------------------- -------------------------------------4 steps, with lora, 100 runssame prompt-------------------------------------total: 4% - strict reflection6% - strict reflection but outside mirror borders21% - wrong symmetry69% - trash (prompt misunderstanding)-------------------------------------Hypothesis: low steps is not enough for generate sophisticated concepts. -------------------------------------8 steps, with lora, 25 runssame prompt -------------------------------------total: 28% - strict reflection16% - strict reflection but outside mirror borders32% - wrong symmetry24% - trash (prompt misunderstanding)------------------------------------- -------------------------------------8 steps, without lora, 25 runssame prompt-------------------------------------total: 16% - strict reflection4% - strict reflection but outside mirror borders24% - wrong symmetry56% - trash (prompt misunderstanding)-------------------------------------Hypothesis: prompt generates too many trash due to "symmetrical" tokens. -------------------------------------8 steps, with lora, 25 runsprompt:-------------------------------------masterpiece, best quality, amazing quality, very aesthetic, absurdres, newest,(safe rating),x_mirror_reflection, {x_reflection_face, {|touching mirror}|x_reflection_facing_aside|x_reflection_facing_away},mirror, reflection, {1girl|1girl|__character__, 2girls}{dynamic pose|lazy pose|sitting}-------------------------------------total: 28% - strict reflection12% - strict reflection but outside mirror borders56% - wrong symmetry4% - trash (prompt misunderstanding)------------------------------------- -------------------------------------8 steps, without lora, 25 runssame prompt-------------------------------------total: 12% - strict reflection0% - strict reflection but outside mirror borders16% - wrong symmetry72% - trash (prompt misunderstanding)-------------------------------------Conclusions: this LoRa version slightly increase probability (+10% for 4 steps, +12-16%) to find strict reflections in a new generation. Increasing steps is refining the results.I go to the next step of the experiment.

Files with SHA256: d40a07c750e9c6629d1c6943f97d8a16e49a7b116de75cf77efb960192135400