BitrateCraft · Source setup · bitratecraft.org

Source, Destination and Output Naming

Everything a transcode does is bracketed by two decisions made before any encoder setting matters: what counts as the source, and where the result will be written. This page covers the scan, title and angle selection, the range modes, the naming template, and what happens when a name is already taken.

HandBrake will not let an operator encode nothing, but it will happily encode the wrong thing. The controls described here exist to make the target explicit, and they are the first thing to check when an output turns out to contain the wrong segment of a longer file.

Scan order

Choosing a file starts a scan. The scan opens the container, enumerates its streams, and reports durations, codecs, dimensions and the title list. Several consequences follow from that order of operations. The scan is the only moment at which the program learns the file's structure, so a value shown afterwards was read from the file and not guessed.

A large or awkwardly structured file may take a noticeable moment to scan, and the panels stay empty until the scan reports back. If a panel looks wrong, rescanning is cheaper than correcting four tabs by hand, because a rescan rebuilds every panel from the file rather than from the previous state.

Choose the source and wait for the scan to complete; an incomplete scan shows a shortened title list.
Confirm duration and dimensions in the summary tab against the material you expect.
Select the title, then set the range, then choose the destination name.
Only after those four steps move on to encoder settings.

Title and angle selection

A title is a distinct playlist inside the container, and a longer file may expose several of them. Titles frequently differ in duration, in the arrangement of their segments, or in the audio tracks they carry. Choosing the wrong title is one of the classic causes of an output that plays a menu, an advertisement reel, or a shortened edit instead of the expected programme.

Angles appear only on material authored with multiple camera perspectives. The angle control picks one and encodes it alone; there is no blended output. Where several titles look similar, duration is the most reliable differentiator, and the summary tab reports it.

Range modes

The range control decides how much of the selected title is encoded. Four modes cover nearly every case. A whole title takes everything from the first frame to the last. A chapter range takes a contiguous set of chapter markers, which is the cleanest way to cut by structure rather than by clock. An explicit start and end accepts frame-accurate times typed by hand, which is the most flexible and the easiest to get wrong.

The fourth mode is the one operators discover last: a range may be expressed as a number of seconds or frames counted from a known point, which suits generated material where chapter markers are absent. Whatever the mode, the resulting duration shown in the summary should be checked against the intended duration before the job is queued, because a mistyped range is invisible in a short test and obvious in a long one.

Destination and output naming

The destination field accepts a path and a file name, and it also accepts a small set of placeholders that are expanded when the job runs. The available tokens typically cover the source name, the title number, the selected preset, the output codec, the date, and a running index. Using a template rather than a fixed name is what makes a batch of thirty files producible without thirty manual renames.

Token	Expands to	Useful when
Source name	The input file name without its extension	One output per source, kept beside its origin
Title number	The selected title index	Several titles are extracted from one file
Preset name	The preset used for the job	Variants of the same source must stay distinguishable
Date	The day the job runs	Deliveries are versioned by date
Index	A counter across the queue	Preventing two jobs from claiming one name

A workable default is the source name followed by the preset and a date. It sorts naturally in a file browser, it keeps every variant visible, and it survives being handed to somebody else who has no access to the queue that produced it.

File conflicts

When the chosen name already exists, the behaviour depends on the mode selected in the same panel. The safe modes refuse to continue, or write the result beside the existing file under a modified name. Neither of those destroys anything, and both make the collision visible.

The risk is the mode that overwrites, and the risk is concentrated in exactly the workflow a template encourages: a batch run twice with the same tokens will collide with its own earlier output on the second pass. Two habits remove the hazard. Extend the template with the date or index token so repeats cannot land on the same name, and point the queue at a working directory that is emptied deliberately rather than at a delivery folder that is read by other people.

A bare hard drive on a bench, representing source and destination storage for a long queue — Source and destination are different jobs for the same disk: read once, write many times, and never to the same path twice in a batch.

Before a long run, confirm three things: that the selected title is the intended one, that the range covers the intended duration, and that the name template cannot collide with an earlier output.

Those three checks take half a minute and remove almost every failure that otherwise appears hours into a queue. The panel tour continues with the settings that follow, and the reference page explains the axes those settings belong to.