invoke-ai/InvokeAI · error · ValueError

denoising_start should be 0 when initial latents are not pro

Error message

denoising_start should be 0 when initial latents are not provided.

What it means

When no initial latents are supplied (pure text-to-video/image), the Wan denoise run must start from pure noise, so denoising_start must be 0. A non-zero denoising_start implies img2img where sampling begins partway along the sigma schedule using init latents; without them it is contradictory, so the node raises. This prevents silently ignoring denoising_start.

Source

Thrown at invokeai/app/invocations/wan_denoise.py:588

            latent_channels=latent_channels,
            height=self.height,
            width=self.width,
            spatial_scale_factor=spatial_scale,
            device=device,
            dtype=latent_dtype,
            seed=self.seed,
        )

        # Combine init latents + noise per the schedule's starting sigma.
        if init_latents_5d is not None:
            if self.add_noise:
                s_0 = float(sigmas[0])
                latents = s_0 * noise + (1.0 - s_0) * init_latents_5d
            else:
                latents = init_latents_5d
        else:
            if self.denoising_start > 1e-5:
                raise ValueError("denoising_start should be 0 when initial latents are not provided.")
            latents = noise

        if total_steps <= 0:
            return latents.squeeze(2)

        # Inpaint extension (4D space — the existing extension is shape-agnostic
        # but operates on the squeezed-T shape we use for masks).
        inpaint_mask = self._prep_inpaint_mask(context, latents.squeeze(2))
        inpaint_extension: RectifiedFlowInpaintExtension | None = None
        if inpaint_mask is not None:
            if init_latents_5d is None:
                raise ValueError("Initial latents are required when using an inpaint mask (img2img inpainting).")
            inpaint_extension = RectifiedFlowInpaintExtension(
                init_latents=init_latents_5d.squeeze(2),
                inpaint_mask=inpaint_mask,
                noise=noise.squeeze(2),
            )

View on GitHub (pinned to 0b6a024f2f)

Solutions

Set denoising_start to 0 (or its default) since no initial latents are provided
Or connect an image via Wan Image to Latents so initial latents exist and denoising_start > 0 is meaningful

Example fix

// before
wanDenoise: denoising_start=0.7, no image input
// after
wanDenoise: denoising_start=0.0  // txt2img
// or: connect image -> wanImageToLatents -> denoise with denoising_start=0.7

Defensive patterns

Strategy: validation

Validate before calling

if init_latents is None and denoising_start > 0:
    raise ValueError("denoising_start requires initial latents; set it to 0 for txt2img")

Try / catch

try:
    result = wan_denoise.invoke(context)
except ValueError as e:
    if 'denoising_start should be 0' in str(e):
        wan_denoise.denoising_start = 0.0
    else:
        raise

Prevention

Only set denoising_start > 0 in img2img workflows
Reset denoising_start to 0 when removing the image input
Use img2img workflow templates rather than editing txt2img ones

When it happens

Trigger: Setting denoising_start > 0 on a WanDenoise node that has no image (and hence no ImageToLatents/initial latents) connected; copying an img2img workflow and disconnecting the image input while leaving denoising_start set.

Common situations: Converting an img2img workflow to txt2img by removing the image input; users experimenting with denoising_start for partial redenoising without realizing it requires init latents.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/3e9e663f1f5c8c06. Report an issue: GitHub.