How to Add Film Grain and Texture to AI Art Without Breaking the Image
AI-generated images have a characteristic smoothness. The rendering is technically perfect in a way that photographs rarely are: no sensor noise, no lens aberration, no film chemistry. For some uses this is fine. For wallpapers and art that should feel tactile, atmospheric, or analog, that smoothness is a problem. It makes the image feel synthetic even when the content is convincing.
Film grain and texture overlays are the most common way to address this, but they have to be applied carefully. Added too heavily, they destroy the fine detail that makes AI art worth using. Added in the wrong way, they create noise that fights the image rather than integrating with it. This article covers the methods that work and the mistakes that do not.
Method 1: Prompting for Texture in Midjourney
The cleanest approach is to build texture into the generation itself rather than adding it in post-processing. When the grain and texture are part of the original output, they integrate naturally with the image content rather than sitting on top of it as a separate layer.
Prompt vocabulary that generates analog texture
- Film grain: "35mm film grain", "analog film texture", "ISO 3200 grain", "heavy film grain", "cinematic grain"
- Photographic texture: "shot on expired film", "film photography", "analog photography aesthetic"
- Painterly and surface texture: "visible brush strokes", "canvas texture", "rough paper grain", "textured surface"
- Print texture: "risograph texture", "screen print texture", "offset print grain", "halftone texture"
- Aged and worn: "aged photograph", "vintage print", "weathered surface texture", "faded analog"
Combining with --stylize for texture amplification
Higher --stylize values push Midjourney's aesthetic processing harder, which tends to amplify stylistic qualities including texture. If your texture prompts are producing images that feel too clean, increasing --stylize from the default 100 to 200-400 often increases the texture and grain presence. The tradeoff is that very high stylize values can also change the composition and subject rendering significantly.
Limitations of the prompt method
Midjourney's grain rendering varies significantly across generations even with the same prompt. Some outputs have rich visible grain; others have very little even with identical prompting. For consistent grain character across a series, post-processing methods give more control.
Method 2: Lightroom Grain
Lightroom's built-in grain tool (Detail > Grain) is the fastest and most practical option for most post-processing grain workflows. It applies luminance-based grain that integrates cleanly into the image at export.
Lightroom grain settings
- Amount: Controls grain intensity. For AI art that should feel analog but not obviously grained, 15-30 is a starting range. For heavy, visible grain that makes a stylistic statement, 40-70. Above 70, grain typically becomes the dominant visual element.
- Size: Controls grain particle size. Small values (10-20) produce fine, photographic grain similar to ISO 400 film. Larger values (30-50) produce the coarser grain of pushed high-ISO film. Larger grain reads as more strongly analog and vintage.
- Roughness: Controls how irregular and organic the grain pattern is. Lower roughness produces uniform, regular grain. Higher roughness (60-100) produces more organic, film-like variation between particles. For analog aesthetics, roughness values of 60-80 are more convincing than low roughness values.
Grain and sharpening interaction
Grain and sharpening work against each other. If you are adding significant grain, reduce or remove sharpening in the Detail panel first. Applying both simultaneously produces a harsh, noisy result rather than organic film grain. For images that need grain for aesthetic purposes, the grain itself provides apparent sharpness through texture contrast.
Masking grain to protect flat areas
One problem with Lightroom's global grain is that it applies evenly to every region of the image, including very smooth, clean areas like sky gradients and out-of-focus backgrounds. Fine grain on a smooth gradient can create visible banding or a slightly structured noise that looks wrong. Using Lightroom's range mask or brush to reduce grain in very smooth regions fixes this.
Method 3: Photoshop Overlay Technique
The Photoshop overlay method uses a separate texture layer blended into the image, giving you significantly more control than Lightroom's grain tool. It is more effort but enables creative control over the type, character, and positioning of texture.
Basic overlay workflow
- Open your image in Photoshop.
- Create a new blank layer above the image layer.
- Fill it with 50% grey (Edit > Fill > 50% Grey).
- Apply Filter > Noise > Add Noise with Gaussian distribution and Monochromatic checked. Amount of 3-8% produces grain; higher values produce more visible noise.
- Set the grey layer blend mode to Overlay or Soft Light.
- Adjust layer opacity to control intensity. 30-60% opacity is a typical starting range.
The Overlay blend mode makes the 50% grey invisible and only lets the light and dark grain particles affect the image. The result is grain that interacts with the image tonally, brightening highlights slightly and darkening shadows, which is closer to how film grain behaves than a flat noise application.
Using real film grain scans
For the most convincing analog texture, use actual scanned film grain overlays rather than digital noise. Film grain overlays are widely available as free and paid assets, and consist of scanned frames of unexposed or minimally exposed film. Placed above your image and blended at 10-30% opacity in Screen, Overlay, or Soft Light mode, they add genuine photochemical texture character that digital noise cannot fully replicate.
Applying texture selectively
Add a layer mask to the grain layer and paint black over any areas where you want no grain (very smooth gradients, important detail areas where grain would obscure fine texture). Selective grain application is more convincing than uniform application across the entire image.