GIF Frame Extractor
A GIF may store only a changed rectangle instead of a full picture for each frame. This tool decodes the compressed pixels and sequentially applies image offsets, transparency and disposal instructions. It produces composited PNGs at the logical screen size and a CSV of the original delays, rather than extracting only the first image.
Key features
- Read all raster frames in GIF87a and GIF89a files and save PNGs
- Handle image offsets, transparency, global/local palettes and interlaced rows
- Apply keep-frame disposal 0/1, background restoration 2 and previous-screen restoration 3
- Export original hundredths-of-a-second delays as unadjusted millisecond CSV
- Download numbered PNGs or a ZIP containing every PNG and timing.csv
- Run an actual GIF-byte example with progress, cancellation and memory limits
How to use
- Choose a GIF file or load the three-frame example.
- Inspect the logical dimensions, frame count and sum of recorded delays.
- Choose transparent clearing or restoration to the GIF logical background color.
- Extract every frame and wait for PNG compositing to finish.
- Inspect the frames and save a selected PNG, timing CSV or the complete ZIP.
Use cases
- Inspect the sequence of movement in an animation as still images
- Check transparent regions and disposal behavior in a web GIF
- Save frame delays and accumulated start times as editing references
- Extract a scene at the GIF’s logical screen dimensions
Frequently asked questions
How does this differ from GIF Maker?
GIF Maker combines still images into an animation and uses only the first frame of an input GIF. This tool does the reverse: it decodes all image frames from one GIF and combines them with preceding content to create full-canvas PNGs.
How are partial images and disposal handled?
A frame can contain only a rectangle of the logical screen. It is drawn at its stored offset, while transparent indices leave existing pixels unchanged. Disposal 0/1 keeps the content, 2 restores that frame rectangle to the selected background, and 3 restores the screen from before the frame. Every PNG has the full logical dimensions.
Can the background choice change the output?
Transparent mode clears the initial screen and disposal-2 regions to transparency. Logical-background mode uses the background color in the GIF global palette, falling back to transparency if no global palette exists. The original webpage background may not be stored in the file, so it cannot be inferred. The preview checkerboard is not saved.
Why can CSV timing differ from playback in a browser?
GIF delays are recorded in hundredths of a second. CSV exports those values multiplied by 10, including zero, and their accumulated start times. Browsers may clamp very short delays or play with additional latency, so this is not a measurement of actual playback. Loop information is also shown as the value stored in the file.
What are the input and output limits?
Input is at most 20 MiB, 2,048px per side and four million screen pixels, with up to 120 frames and at most 40 million screen-pixels × frames. PNG output totals at most 64 MiB. Device memory can impose lower limits. Plain Text Extensions, user-input GIFs, reserved disposal values and damaged data are explicitly rejected; a partial extraction is never reported as complete.
How do cancellation, pixel aspect ratio and downloads work?
Cancellation discards pending output. If PNG encoding or ZIP generation has started, the next job waits for its callback to settle. Non-square pixel aspect ratios are reported without stretching the PNG. The ZIP includes numbered PNGs, timing.csv and a background-policy note. Actual saving depends on browser download settings.
Privacy
GIF files and extracted frames are processed in current browser memory without upload or automatic persistence by this tool. External addresses in a GIF are not opened or fetched. Replacing a file, changing settings, clearing or leaving the page releases old result URLs. Downloads save the generated PNG, CSV or ZIP to your device.
Comments & questions