invoke-ai/InvokeAI · error · ValueError

Initial latents are required when using an inpaint mask (ima

Error message

Initial latents are required when using an inpaint mask (image-to-image inpainting)

What it means

An inpaint mask requires the original image's latents so the extension can composite inpainted regions back against known content. If a mask is present but init_latents is None, InvokeAI raises this ValueError because inpainting is impossible without reference latents.

Source

Thrown at invokeai/app/invocations/anima_denoise.py:640

                s_0 = sigmas[0]
                latents = s_0 * noise + (1.0 - s_0) * init_latents
            else:
                latents = init_latents
        else:
            if self.denoising_start > 1e-5:
                raise ValueError("denoising_start should be 0 when initial latents are not provided.")
            assert noise is not None
            latents = noise

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

        # Prepare inpaint extension
        inpaint_mask = self._prep_inpaint_mask(context, latents.squeeze(2))
        inpaint_extension: AnimaInpaintExtension | None = None
        if inpaint_mask is not None:
            if init_latents is None:
                raise ValueError("Initial latents are required when using an inpaint mask (image-to-image inpainting)")
            assert noise is not None
            inpaint_extension = AnimaInpaintExtension(
                init_latents=init_latents.squeeze(2),
                inpaint_mask=inpaint_mask,
                noise=noise.squeeze(2),
                shift=ANIMA_SHIFT,
            )

        step_callback = self._build_step_callback(context)

        # Initialize scheduler driver if not using built-in Euler.
        use_scheduler = self.scheduler != "euler"
        driver: AnimaSchedulerDriver | None = None
        if use_scheduler:
            driver = AnimaSchedulerDriver(
                scheduler_name=self.scheduler,
                sigmas=sigmas,
                steps=self.steps,

View on GitHub (pinned to 0b6a024f2f)

Solutions

  1. Provide an init image so init_latents can be encoded (img2img inpainting).
  2. Remove or disconnect the inpaint mask if you truly want unconditional txt2img generation.
  3. Verify upstream nodes actually emit the init image/latents into the denoise node.

Example fix

// before
denoise.mask = mask_field  # no image input
// after
denoise.image = init_image_field  # img2img inpainting
// or remove the mask for txt2img
Defensive patterns

Strategy: validation

Validate before calling

if inpaint_mask is not None and init_latents is None:
    raise ValueError("Inpainting requires an init image/latents")

Try / catch

try:
    output = invoker.invoke(denoise_invocation)
except ValueError as e:
    if "inpaint mask" in str(e):
        raise RuntimeError("Connect an init image or remove the mask before running") from e

Prevention

When it happens

Trigger: Supplying an inpaint_mask to the Anima denoise invocation (mask detected by _prep_inpaint_mask) while running in pure txt2img mode with no initial latents.

Common situations: Using a canvas/mask workflow where the init image layer was removed or never rendered; thinking a mask alone is enough for inpainting; converting an img2img inpaint workflow to txt2img but keeping the mask.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/cf47d95337ddc12e. Report an issue: GitHub.