Technique

Generational Loss

Generational Loss

Glitch art that uses generational loss and compression artifacts exploits how digital media degrades when it is repeatedly saved, exported, or re-encoded. By pushing codecs beyond their intended use, you can sculpt blocky, smeared, and noisy errors into a deliberate aesthetic.

1. What are generational loss and compression artifacts?

Generational loss
In digital media, generational loss is the gradual degradation that occurs when you re-save or re-encode a file multiple times, especially with lossy formats like JPEG, MP3, or H.264. Each “generation” throws away data the algorithm considers non-essential, compounding small losses into visible breakdown.

Compression artifacts
Lossy codecs reduce file size by approximating visual or audio information. Their shortcuts create characteristic artifacts:

  • Blockiness and macroblocking
  • Color banding and washed gradients
  • Mosquito noise around edges
  • Smearing or ghosting in motion video
  • Ringing/halos along high-contrast lines

Glitch artists deliberately emphasize these “errors” instead of hiding them.

2. How lossy compression works (in artist terms)

Different codecs vary, but most image/video lossy compression follows a similar logic:

  1. Transform
    The image is split into blocks (often 8×8 pixels) and transformed into frequency data. Flat areas are “low frequency”; edges and fine details are “high frequency.”

  2. Quantization
    High-frequency data is aggressively simplified or discarded. The more you compress, the more detail is lost.

  3. Encoding
    The simplified data is stored efficiently so it can be decoded back into an image.

When you re-save with lossy compression:

  • Previously discarded detail cannot be recovered.
  • New approximations are made on already-damaged data.
  • Blocks, bands, and noise become more pronounced and stylized.

Glitch art leans into this cumulative breakdown as a visual language.

3. Core techniques using generational loss

A. JPEG crushing (images)

This is a classic approach for still images:

  1. Start with a high-resolution clean image.
  2. Save as JPEG with a very low quality setting (for example, 0–20%).
  3. Reopen that file, save again at low quality.
  4. Repeat this 5–30 times, watching the breakdown evolve.

Variations:

  • Alternate between high and low quality to create uneven artifact patterns.
  • Upscale the degraded image, then crush it again to exaggerate block size.
  • Change color space (sRGB to Adobe RGB and back) between saves to introduce subtle shifts.

B. Video recompression loops

For moving image glitch:

  1. Export a video as H.264 with low bitrate and high compression.
  2. Import the exported file back into your editor.
  3. Re-export with the same or worse settings.
  4. Repeat until motion smears, blocks “stick,” and details melt into abstraction.

Variations:

  • Change aspect ratio each generation (16:9 → 4:3 → 9:16) to warp shapes.
  • Alternate codecs (H.264 → HEVC → VP9 → H.264) to layer artifact “dialects.”
  • Add simple edits (cuts, speed changes) between generations so errors don’t become too uniform.

C. Hybrid databending + compression

Combine file corruption with normal compression:

  • Open a JPEG or MP4 in a hex editor or text editor (after changing extension in some cases).
  • Randomly copy, delete, or paste chunks of data, avoiding the very beginning (header) so the file remains readable.
  • Save, then open and resave repeatedly in a normal editor to “smooth” and spread the corruption via compression.

This yields more chaotic, non-standard artifacts that compression then stabilizes into a new texture.

4. Practical tips for control and experimentation

1. Work non-destructively

  • Always keep a clean master.
  • Run generational experiments on duplicated files or folders.
  • Save intermediate generations that look promising so you can branch off.

2. Use batch processing

Tools like ImageMagick, FFmpeg, or actions/scripts in Photoshop and similar editors can automate:

  • Re-saving JPEGs with specific quality levels.
  • Re-encoding video multiple times with given bitrates.
  • Renaming generations (image_01.jpg, image_02.jpg, etc.).

Automation lets you explore deeper degradation than you would manually.

3. Push codecs to their breaking point

  • Use very low bitrates or quality settings.
  • Crank down resolution, then upscale again.
  • Encode at unusual frame rates or aspect ratios for stranger distortion.

Expect crashes, unreadable files, and happy accidents - that instability is part of the practice.

4. Think compositionally

Generational loss is not just a filter; treat it like paint:

  • Reserve strong artifacts for focal areas by selectively recompressing regions (via masks or layer compositing).
  • Blend clean and broken versions with layer modes (Overlay, Difference, Screen) to control intensity.
  • Stack multiple glitch passes: color shift pass, then JPEG crush, then subtle noise.

5. Archive and document

  • Note settings: codec, bitrate, quality level, number of generations.
  • Save side-by-side comparisons: original vs gen 5 vs gen 15.
  • Build your own “codec library” of favorite settings as a personal glitch toolkit.

5. Conceptual considerations

Generational-loss glitch is not only an effect; it can speak about:

  • Memory and decay in digital culture
  • Media circulation and over-sharing
  • Compression of identity and information under platform constraints

Leveraging these processes consciously can give your work depth beyond visual novelty.

By understanding how compression fails, you gain creative control over a family of glitches that once seemed purely accidental.