<pre><code>Based on some renders of Lara Croft of CORE design. only official TR 1-5 PS, PC, high resolution 3D artwork and some video frames.

Other version:

https://civarchive.com/models/209169

difference in training steps, when lora created with CivitAI it give 10 versions, this is 10, It is selected by me as good result.

Promts: lara croft, front view, back view, side view, close up, portait, view from up, lying, sitting, standing, standing in shadows,

Example settings:
Sampler DPM++ 2M Karras CFG scale 7.5 Steps 56

{
  "unetLR": 0.0005,
  "clipSkip": 1,
  "loraType": "lora",
  "keepTokens": 0,
  "networkDim": 32,
  "numRepeats": 5,
  "resolution": 1024,
  "lrScheduler": "cosine_with_restarts",
  "minSnrGamma": 5,
  "targetSteps": 350,
  "enableBucket": true,
  "networkAlpha": 16,
  "optimizerArgs": "weight_decay=0.1",
  "optimizerType": "AdamW8Bit",
  "textEncoderLR": 0.00005,
  "maxTrainEpochs": 10,
  "shuffleCaption": false,
  "trainBatchSize": 6,
  "flipAugmentation": false,
  "lrSchedulerNumCycles": 3
}</code></pre>

Lara Croft

<pre><code>Based on some renders of Lara Croft of CORE design. only official TR 1-5 PS, PC, high resolution 3D artwork and some video frames.

Other version:

https://civitai.com/models/209169

difference in training steps, when lora created with CivitAI it give 10 versions, this is 6, It is selected by me as good result.


Promts: lara croft, side view, front view, back view, sitting, standing, standing in shadows, lying, portait, view from up, close up,
{
  "unetLR": 0.0005,
  "clipSkip": 1,
  "loraType": "lora",
  "keepTokens": 0,
  "networkDim": 32,
  "numRepeats": 5,
  "resolution": 1024,
  "lrScheduler": "cosine_with_restarts",
  "minSnrGamma": 5,
  "targetSteps": 350,
  "enableBucket": true,
  "networkAlpha": 16,
  "optimizerArgs": "weight_decay=0.1",
  "optimizerType": "AdamW8Bit",
  "textEncoderLR": 0.00005,
  "maxTrainEpochs": 10,
  "shuffleCaption": false,
  "trainBatchSize": 6,
  "flipAugmentation": false,
  "lrSchedulerNumCycles": 3
}</code></pre>

Files with SHA256: 9252f5b3b7103fc04b2d8736f91b33dfe77ca9e8c5dfd7722cbcce82a3032776