ComfyUI Node
ROCm Text Generate LTX2 Prompt
ROCm-friendly LTX2 prompt generation. Same I/O as TextGenerateLTX2Prompt. For best speed on ROCm run ComfyUI with: --use-pytorch-attention
ROCm Text Generate LTX2 Prompt
- clip
- image
- generated_text
◄prompt►
◄max_length256►
◄sampling_modeon►
◄temperature0.7000►
◄top_k64►
◄top_p0.95►
◄min_p0.05►
◄repetition_penalty1.05►
◄seed0►
CategoryROCm Ninodes/Generative AI
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | CLIP/LLM model (e.g. Gemma3 12B for LTX2). | |
| prompt | STRING | User raw input prompt to expand into a video generation prompt. | |
| max_length | INT | 2561–2048 | Maximum number of tokens to generate. |
| imageopt | IMAGE | Optional first frame for image-to-video (I2V) mode. | |
| sampling_modeopt | COMBO | on | Sampling on = stochastic (temperature, top_k, top_p); off = greedy. |
| temperatureopt | FLOAT | 0.70000.01–2 | Sampling temperature when sampling_mode is on. |
| top_kopt | INT | 640–1000 | — |
| top_popt | FLOAT | 0.950–1 | — |
| min_popt | FLOAT | 0.050–1 | — |
| repetition_penaltyopt | FLOAT | 1.050–5 | — |
| seedopt | INT | 00–18446744073709550000 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| generated_text | STRING | — |