Old Photo Restoration

Actual generation · grok-imagine-image-2.0 · 4:3 · 1K · AI-generated, not a real photograph

Download image for free

AI Photography

Old Photo Restoration

AI Photography images and AI prompts, organized by subject and creative use.

Suggested canvas 4:3 Suggested model · Grok Imagine 2.0 English prompt

Full prompt

Restore the source photograph while strictly preserving facial geometry, body proportions, identity, original aspect ratio, framing, semantic positioning, and all existing objects. Target identity similarity ≥ 0.95. Reconstruct natural macro epidermal texture, visible skin pores, textile weaves, and era-specific surface details; avoid "plastic" smoothing. First assess whether the source is real, synthetic, or ambiguous, and estimate overall damage on a 0–100 scale with regional confidence (RC). For RC ≥ 70, use context-guided high-detail reconstruction. For RC 30–69, perform coarse-to-medium structural restoration with zero semantic invention. For RC < 30, use only era-consistent silhouette and texture filling. Apply area-based reconstruction constraints: under 10% damage uses guided inpainting, 10–40% uses structural geometry synthesis, and over 40% uses a texture-only pass. Never invent, remove, or alter meaningful objects or identifying marks. Maintain rigorous vanishing-point alignment, scale, geometric continuity, and physical plausibility. Validate reconstructed areas against surrounding geometry and revert to "Safe Fill" if the physics check fails. Restore chiaroscuro-level lighting depth while preserving the original shadow direction and luminance gradients. Reconstruct physically accurate material reflectance, including appropriate specular highlights and matte finishes. Recover faded colors with an era-appropriate palette and controlled HDR tonal range, retaining vintage warmth, analog character, and spectral color fidelity. Apply sub-pixel deconvolution, motion-blur neutralization, chromatic-aberration correction, and multi-stage 4x-or-greater super-resolution with anti-artifact refinement. Separate sensor noise from compression artifacts, remove degradational "noise.", and preserve the original analog grain structure. Preserve metadata and EXIF information when possible. Intended output: lossless 16-bit PNG or TIFF in archival and sRGB color spaces, with no watermarks, logos, or altered identifying marks.

Share
Create in the workspace

How to adapt this brief

AI Photography images and AI prompts, organized by subject and creative use.

Browsing and copying are free. Generation uses account credits with a quote before submission. Model, language and randomness affect output; exact reproduction is not guaranteed.