TikTok Quality Optimizer. AST Engine
Binary AST pipeline — FFmpeg faststart remux + fake sample injection + chunk offset shifting. Zero re-encoding, maximum quality bypass.
lock System Authentication
Enter your Haze license key to unlock optimization capabilities and higher limits.
Join discord.gg/hazedf, go to #bot-cmds and use /sign-in to get your license.
User
Haze Method Protocol v8.0 AST ENGINE
Full binary AST patcher — restructures MP4 container via FFmpeg faststart, injects fake video samples into chunk tables, and shifts all offsets to bypass platform compression.
AST Binary Patcher
This method uses the full Haze AST engine — FFmpeg faststart restructure + binary chunk injection. It injects fake video samples into your MP4's sample tables and shifts all chunk offsets, making the platform treat your file as already optimized.
Click or drag video to optimize
AST Engine Active · H.264 / HEVC · MP4 / MOV
How the Haze AST Engine Works
AST EngineThe Haze AST Engine uses a multi-stage binary pipeline to restructure your MP4 container at the atom level. It first runs an FFmpeg faststart remux to normalize the container, then applies a custom Binary AST Patcher that manipulates the video's sample tables and chunk offset maps.
When platforms like TikTok receive a video, their automated transcoding pipelines analyze the container structure to decide how to process it. By injecting carefully calculated fake samples into the video track's stsz, stsc, and stco/co64 tables, the AST Engine makes the platform's ingestion server believe the video has already been processed and optimized, causing it to skip destructive re-compression.
The AST Pipeline
The AST Engine runs a 3-stage pipeline entirely in your browser using WebAssembly:
Stage 1: ffmpeg -c copy -movflags +faststart -map_metadata -1
Stage 2: AST Parse → stsz/stsc/stco injection (fake samples)
Stage 3: Multi-pass offset recalculation + mdat assembly
- Stage 1 — FFmpeg Faststart Remux: Losslessly copies all streams into a clean MP4 container with
moovatom moved beforemdat. Strips all metadata and chapter markers for a pristine container. - Stage 2 — Binary AST Patcher: Recursively parses the MP4 atom tree (
moov → trak → mdia → minf → stbl), identifies the video track, and injects thousands of fake 8-byte samples into thestsz(sample size),stsc(sample-to-chunk), andstco/co64(chunk offset) tables. - Stage 3 — Offset Recalculation: Runs multiple passes to recalculate all absolute byte offsets in the container, ensuring every chunk pointer remains valid after the moov size changes from the injection.
- Handler Normalization: Patches
hdlrnames toVideoHandler/SoundHandlerand normalizesmdhdlanguage codes to match standard camera output signatures.
Frequently Asked Questions
How does this bypass TikTok compression?
The AST Engine injects fake sample entries into the video track's chunk tables. When TikTok's server analyzes the container, the inflated sample count and offset structure make it believe the video has already been transcoded to an optimal format, so it skips its own re-compression pipeline and serves your original quality.
Does this re-encode my video?
No. The entire pipeline uses -c copy (stream copy). Your original video and audio bitstreams are preserved byte-for-byte. Only the container metadata is modified. This means zero quality loss and processing takes seconds, not minutes.
Is this tool safe and private?
Yes. All processing happens entirely on your local device using client-side JavaScript and WebAssembly FFmpeg. Your video files are never uploaded to any server. The source code is fully visible in your browser's developer tools.
Which codecs and formats are supported?
The tool supports H.264 (AVC) and HEVC (H.265) in MP4 and MOV containers. The AST patcher works at the container level, so it's codec-agnostic — it manipulates the MP4 atom structure regardless of the underlying video codec.
What's the recommended input?
For best results, use 1080p or higher resolution at 60FPS with a bitrate of 8-20 Mbps. The engine preserves your original quality, so the better your source file, the better the output on TikTok.