COMPARE · CONSTRAIN · SELECT
Synthetic bitrate ladder: choose feasible renditions
Change complexity, desired quality and bitrate budget. See which resolution can meet each target—and why others cannot.
Selected ladder
For each desired quality level, choose the lowest bitrate that reaches it among the seven curves and respects the budget. Missing rungs stay visibly infeasible; they never fall back to a misleading low bitrate.
Your experiment
Start at quality 80. Move complexity from 0 to 100: required rates triple under this formula, while the quality ceilings stay fixed. Then reduce the bitrate budget until the highest rung disappears.
Set target quality to 100. No finite bitrate reaches an asymptotic ceiling of 100, so the final target is correctly marked unreachable. Lower-quality rungs may still be feasible.
Candidates for the highest target
| Resolution | Synthetic ceiling | Required bitrate | Feasibility |
|---|
Inspect the exact model
Each curve uses quality = ceiling × (1 − exp(−bitrate / scale)), where scale = (height / 240) × 0.4 × (0.5 + complexity / 100). Inverting gives required bitrate = −scale × ln(1 − target / ceiling), defined only when target is below the ceiling.
The desired levels are 25, 40, 55, 70, 85 and 95 below the chosen target, plus the target itself. Rungs are selected independently; the same resolution can occur at different bitrates. Dots mark selected feasible rungs. The dashed lines show the final target and maximum allowed bitrate.
Full-precision probe inspector
The 60-second playback is inspection time, not encoding latency. Pause and change speed to inspect a point; Restart keeps your parameters. All curves, ceilings and complexity scaling are invented teaching assumptions. This page does not encode video, compute VMAF, measure viewer experience, construct a production convex hull or estimate bandwidth savings. Real per-title ladders require measured encode candidates and device/content constraints; these synthetic choices must not be used as production encoder settings.
Choose under explicit synthetic assumptions
Seven analytical bitrate-quality curves represent heights from two hundred forty through two thousand one hundred sixty pixels. Every curve has an invented ceiling and scale; it is not derived from an encoded video or a perceptual benchmark. The model computes quality as ceiling multiplied by one minus an exponential decay. It analytically inverts that expression to find the rate needed for a target, then selects the lowest feasible rate among the seven candidates. This makes the optimization inspectable without implying measured codec efficiency.
Understand each model control
Synthetic complexity ranges from zero through one hundred and changes the horizontal scale of every curve. Highest desired quality sets the final target and the lower rung targets included in the ladder. Maximum rendition bitrate is a per-rendition upper bound, not a total bandwidth or storage budget across the entire ladder. Inputs immediately rebuild the curves, candidate table and choices, and reset inspection to zero. Restore defaults restores complexity fifty, target eighty and budget twenty megabits per second. A higher scale increases required rates without changing any quality ceiling.
Reproduce the complexity multiplier
Keep target eighty and a generous bitrate budget. Compare complexity zero with one hundred. The scale factor changes from one half to one and a half, so the required bitrate of every reachable candidate triples under this formula. The selected resolution stays the same when both cases remain feasible. Read the full candidate table to verify the relationship rather than comparing rounded plot positions. Reduce the budget to exclude the preferred choice: the result must report infeasibility when no candidate can reach the target inside the bound. A budget limit never creates a lower-rate fallback that falsely meets quality.
Explore ceilings and independent rungs
Set target quality to one hundred. Every ceiling is at most one hundred, and this asymptotic formula reaches a ceiling only at infinite bitrate, so no finite candidate can meet that target. Lower ladder rungs can still be feasible. Desired levels come from twenty-five, forty, fifty-five, seventy, eighty-five and ninety-five below the chosen target, plus that target. Each level is optimized independently. The same resolution may therefore appear at multiple bitrates; the output is not a production convex hull, a codec switching set or a guarantee of uniform perceptual spacing.
Use the probe to inspect evaluated points
Play moves a probe across each curve from zero to the selected bitrate budget over a sixty-second inspection. Play toggles to Pause; Next curve advances to the beginning of the next curve, and Restart rewinds without changing model controls. Inspection scrubs to a chosen time and pauses. Speed offers quarter, half, normal and double pace without changing the formula. At defaults, scrub to thirty seconds: the probe is on the seven-hundred-twenty-pixel curve at ten megabits per second. The full-precision inspector provides bitrate, computed quality, required rate and feasibility independently of canvas color.
Know when to replace assumptions with measurements
The candidate table and selected-rung list expose all seven choices and their reasons. Quality ceiling means no finite solution; over bitrate budget means a solution exists but violates the current constraint. The sixty-second animation is viewing time, not encoding latency. The model excludes source motion, codec and preset, scene diversity, device decoding limits, audio, storage, packaging overhead and measured VMAF or PSNR. Its synthetic complexity control cannot predict savings for a real title. A production decision needs actual encoded candidates, an explicit quality metric and practical delivery constraints; this lab only demonstrates bounded selection on the stated curves.