---
library_name: diffusers
pipeline_tag: image-to-image
tags:
- seismic-inversion
- impedance-inversion
- diffusion
- ddpm
- overthrust
---

# Seismic-LDDPM

Seismic-LDDPM is a latent DDPM pipeline for seismic impedance inversion. The
pipeline takes a low-frequency impedance image (`dipin`) and a synthetic seismic
record (`record`) and predicts the impedance image.

This repository includes:

- Diffusers-format model components: `vq_model`, `unet`, `scheduler`, and
  `condition_encoder`.
- `SeismicImpInvLDDPMPipeline` in `codes/pipeline.py`.
- A complete Overthrust benchmark sample at `data/Overthrust_trueimp.mat`.
- Root `infer.py` and supporting inference code under `codes/`.

## Installation

```bash
git clone https://huggingface.co/mally-2000/saii-cldm-synthetic
cd saii-cldm-synthetic
pip install -r requirements.txt
```

## Overthrust Evaluation

The Overthrust evaluation script is intentionally fixed to the bundled
`data/Overthrust_trueimp.mat`. It cuts the full model into six `256 x 256`
patches, synthesizes the seismic records and low-frequency impedance inputs,
runs inference, stitches the six predictions back together, and computes the
metrics.

```bash
python codes/eval_overthrust.py \
  --model . \
  --output outputs/overthrust \
  --num-inference-steps 1000
```

Outputs:

- `outputs/overthrust/full_target.npy`
- `outputs/overthrust/full_prediction.npy`
- `outputs/overthrust/full_reconstruction.npy`
- `outputs/overthrust/comparison_impedance.png`
- `outputs/overthrust/metrics_summary.json`

## Benchmark Result

Evaluated locally on the bundled Overthrust benchmark with 1000 DDPM steps,
`noise_snr=15`, `dipin_v=0.012`, `f0=30`, `phase=0`, `seed=1234`, and patch
indices `[0, 1, 2, 3, 4, 5]`.

| Space | PSNR | SSIM | PCC | RRE | NMSE |
|---|---:|---:|---:|---:|---:|
| Normalized | 30.7698 | 0.9339 | 0.9963 | 0.0435 | 0.001894 |
| Impedance | 33.4413 | 0.9554 | 0.9957 | 0.0324 | 0.001050 |
| VQ reconstruction | 37.7954 | 0.9677 | 0.9983 | 0.0209 | 0.000435 |

![Overthrust evaluation](assets/demo.png)

## Single-Sample Inference

For a single default Overthrust patch:

```bash
python infer.py
```

The script builds one Overthrust test sample internally, synthesizes the
low-frequency impedance and seismic record, and saves `prediction.npy`,
`target.npy`, and `comparison.png` under `outputs/infer_LDDPM`.

For SAII-CLDM model-driven sampling:

```bash
python infer.py CLDM
```

## Python Usage

```python
import torch
from codes.pipeline import SeismicImpInvLDDPMPipeline

pipe = SeismicImpInvLDDPMPipeline.from_pretrained(
    "mally-2000/saii-cldm-synthetic",
    torch_dtype=torch.float32,
    trust_remote_code=True,
).to("cuda")

result = pipe(
    dipin=dipin,      # torch.Tensor, BCHW
    record=record,    # torch.Tensor, BCHW
    num_inference_steps=1000,
    seed=1234,
)

prediction = result.impedance_samples
```

## Notes

- `codes/dataset.py` contains a lightweight `SeismicBase` and
  `OverthrustTrueimpDataset`; it does not depend on the original training
  repository's `ldm.data.seisimic`.
- Synthetic record generation is seeded through the benchmark configuration so
  the published Overthrust evaluation is reproducible.
- The bundled Overthrust file is used only as a compact benchmark input for
  reproducing this model's inference pipeline.