What happens if a generation fails?

Answer

Short answer:

If a generation fails due to a technical error, check your internet connection, ensure your input file is under 16MB and your inputs does not contain something might trigger content moderation filters. If the failure is visual, meaning the output is inaccurate or contains artifacts, this is part of the iteration process and can often be fixed by refining the prompt, using editing tools, or adjusting the base image.

Overview

In architectural visualization using generative AI, a "failure" generally falls into one of two categories: technical interruptions or visual inaccuracies. Rendair AI is designed to minimize friction, but understanding the difference between these two scenarios helps you resolve the issue quickly.

Technical failures prevent the image from being created at all, often resulting in an error message or a timed-out request. Visual failures occur when the image is generated successfully but does not meet professional standards or architectural intent.

Troubleshooting technical failures

If the system fails to produce an image, it is usually due to connectivity or input constraints.

• Check your connection: Rendair AI is entirely web-based and does not support offline usage. A stable internet connection is required to communicate with the rendering servers.

• Verify file size: The maximum upload size for base images is 16MB. If your input file exceeds this, the generation will not process.

• Review file formats: Ensure you are using supported formats (JPEG, PNG, WEBP, TIFF, HEIC, AVIF). If you are using a PDF, convert it to a JPG before uploading.

• Queue status: Users on the AI Creator plan may experience slower processing times due to queueing. If a generation seems stuck, it may simply be in the queue. Pro and Team Pro plans skip these queues.

Addressing visual inaccuracies

Generative AI is probabilistic, meaning it interprets inputs rather than calculating them physically. If the system generates an image that is stylistically correct but geometrically inaccurate, use the following workflows to correct it.

• Refine the prompt: If the AI misunderstood a material or atmosphere, clarify the text instructions.

• Use editing tools: Instead of regenerating the entire image, use the Inpaint or Selection tools (brush/lasso) to fix specific areas. You can remove unwanted elements or add missing context without altering the rest of the image.

• Adjust the base input: If the AI struggles with perspective, ensure your base 3D screenshot or sketch has clear lines. For 3D workflows, a textured screenshot often provides better guidance than an untextured "clay" mode.

• Check resolution: Default images are approximately 1MP (or 2K for specific models like Nano Banana). If the image lacks detail, use the Upscale tool to increase resolution to 4K or higher, which often clarifies textures and edges.

Support and assistance

If you encounter persistent failures that standard troubleshooting does not resolve, support channels are available based on your subscription tier.

• General Support: Issues are handled by requesting input files and clarifying goals to identify why a specific generation is failing.

• Pro and Team Pro: These users have access to priority support and can book private training sessions. This is often the fastest way to resolve complex workflow failures or learn how to prompt for specific architectural outcomes.