invoke-ai/InvokeAI · error · ValueError

fill_conditioning was provided, but the model is not a FLUX

Error message

fill_conditioning was provided, but the model is not a FLUX Fill model.

What it means

fill_conditioning (e.g. an inpaint mask/image used for FLUX Fill inpainting) is only consumed by FLUX Fill (DevFill variant) transformers. If the loaded transformer is not the DevFill variant, providing fill_conditioning is a configuration mistake, so the invocation raises instead of silently ignoring the conditioning.

Source

Thrown at invokeai/app/invocations/flux_denoise.py:354

            assert noise is not None
            x = noise

        # If len(timesteps) == 1, then short-circuit. We are just noising the input latents, but not taking any
        # denoising steps.
        if len(timesteps) <= 1:
            return x

        if is_schnell and self.control_lora:
            raise ValueError("Control LoRAs cannot be used with FLUX Schnell")

        # Prepare the extra image conditioning tensor (img_cond) for either FLUX structural control or FLUX Fill.
        img_cond: torch.Tensor | None = None
        is_flux_fill = transformer_config.variant is FluxVariantType.DevFill
        if is_flux_fill:
            img_cond = self._prep_flux_fill_img_cond(context, device=device, dtype=inference_dtype)
        else:
            if self.fill_conditioning is not None:
                raise ValueError("fill_conditioning was provided, but the model is not a FLUX Fill model.")

            if self.control_lora is not None:
                img_cond = self._prep_structural_control_img_cond(context)

        inpaint_mask = self._prep_inpaint_mask(context, x)

        img_ids = generate_img_ids(h=latent_h, w=latent_w, batch_size=b, device=x.device, dtype=x.dtype)

        # Pack all latent tensors.
        init_latents = pack(init_latents) if init_latents is not None else None
        inpaint_mask = pack(inpaint_mask) if inpaint_mask is not None else None
        noise = pack(noise)
        x = pack(x)

        # Now that we have 'packed' the latent tensors, verify that we calculated the image_seq_len, packed_h, and
        # packed_w correctly.
        assert packed_h * packed_w == x.shape[1]

View on GitHub (pinned to 0b6a024f2f)

Solutions

Load/select a FLUX Fill (DevFill variant) model in the model loader feeding the denoise invocation.
Remove the fill_conditioning input if inpainting via FLUX Fill is not intended (use the inpaint_mask field for regular masking instead).
Verify the transformer config variant is FluxVariantType.DevFill before connecting fill_conditioning.

Example fix

// before
denoise.fill_conditioning = fillMaskField; // model is plain FLUX Dev
// after
denoise.fill_conditioning = null; // or load a FLUX Fill model instead

Defensive patterns

Strategy: validation

Validate before calling

if denoise.fill_conditioning is not None and model_config.variant != FluxVariantType.DevFill:
    raise ValueError("fill_conditioning requires a FLUX Fill (DevFill) model")

Type guard

def supports_fill_conditioning(config) -> bool:
    return getattr(config, 'variant', None) == FluxVariantType.DevFill

Try / catch

try:
    result = invoke(denoise)
except ValueError as e:
    if 'not a FLUX Fill model' in str(e):
        denoise.fill_conditioning = None  # fall back to no fill conditioning
        result = invoke(denoise)
    else:
        raise

Prevention

Load a DevFill-variant model whenever fill_conditioning is wired
Use inpaint_mask for generic inpainting instead of fill_conditioning on non-Fill models
Check transformer config.variant before connecting conditioning inputs

When it happens

Trigger: Passing a non-null fill_conditioning field to FLUX Denoise while the loaded transformer's config.variant is any FluxVariantType other than DevFill (e.g. Dev, Schnell, Kontext).

Common situations: User wires an inpaint/fill conditioning input but the model selector points at a regular FLUX Dev checkpoint instead of a FLUX Fill model; model was upgraded/replaced and the fill model is no longer loaded; workflow copied from a FLUX Fill example with a different model chosen.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/1891c5cdf3780ae2. Report an issue: GitHub.