after encode_model ,decode failed

[maguslin]

all decode_latent_images are black,latent = xm.encoder.encode_to_bottleneck(batch)

render_mode = 'stf' # you can change this to 'nerf'
size = 128    # recommended that you lower resolution when using nerf

# print("latent shape =", latent.shape)
# print("latent mean =", latent.abs().mean().item())
# print("latent max =", latent.abs().max().item())
# print("xm.renderer=", xm.renderer)

cameras = create_pan_cameras(size, device)
images = decode_latent_images(xm, latent, cameras, rendering_mode=render_mode)
os.makedirs("render_frames", exist_ok=True)
for idx, img in enumerate(images):
    img.save(f"render_frames/frame_{idx:03d}.png")   use shap_e/examples/example_data/cactus/object.obj

[Ribe1414]

Ich danke an alles, aber ich haben keine Ahnung wie das machen