Encrust
Resin printing, file formats and how a slicer works

File formats

What your printer ignores in the file you sliced

6 min read

Resin printer firmware matches the machine name written in a sliced file, and no file format says whether it obeys per-layer settings or variable layer height. Of the 199 profiles Encrust ships, 32 turn per-layer settings off and none claims variable height. In one print on our Saturn 4 Ultra, a file with the flag on stalled the plate; with it off the same job finished.

Thirty-two of the 199 printer profiles Encrust ships mark the machine as not reading per-layer settings. Twenty-six say so because the maker’s published profile asks for a .ctb older than version 4. Three say so because of how the machine moves: the ELEGOO Saturn 4 Ultra, its 16K sibling and the Mars 5 Ultra separate each layer by tilting the vat instead of lifting the plate. The last three, the Anycubic Photon S and two PioNext machines, have the flag off in their profiles with no reason recorded.

In one print on our Saturn 4 Ultra, a file with the flag on stopped the plate rising at about 0.6 mm while the panel kept exposing; with the flag off, the same job finished. The rest of this post is that flag, its sibling for layer thickness, and the name string that decides whether a printer takes the file at all. Everything below about a machine’s behaviour is what the Encrust notes record, and the maker’s manual comes first.

The machine name the file carries

A sliced file names the machine it is for, and the firmware matches that string (profiles, “What the file calls the machine”). The string is not the one a slicer’s picker shows. Encrust lists Saturn 4 Ultra; a file the vendor’s own slicer wrote for that machine carries ELEGOO Saturn 4 Ultra.

You cannot build the second from the first. Joining the manufacturer and the model gives Elegoo Saturn 4 Ultra, and Elegoo is not ELEGOO. So a profile can state the exact string in a machine_name key, and the writers use it, falling back to the display name when it is absent.

Where the string lands depends on the container:

Container Where the name goes Source
.goo two 32-byte fields at offset 92, machine name and machine type, both the same string goo notes
.ctb the slicer-info block, and in version 5 the resin-parameters block too; not padded to a fixed width Chitu notes
.cxdlp not the whole name: a CL- or CT- model code is picked out of it Creality notes

For Creality, Encrust refuses to write at all when the name holds no such code, rather than guess one. For the other containers it writes whatever string the profile gives.

Here is the gap. Of the 199 shipped profiles, 174 set machine_name, the string the maker’s own slicer writes; the Saturn 4 Ultra’s was read out of a file the vendor’s slicer wrote for it. The other 25 write their display name, which may differ from what the firmware matches. What a printer’s screen says when it declines the file, we have not recorded.

If your printer rejects the file by name, the profile guide says where to set machine_name. Use the string your maker’s own slicer writes.

Per-layer settings

Per-layer settings mean the machine reads exposure and motion from a table in front of each layer, instead of from the header alone. In .goo the same idea is called advance mode, and it is one byte at offset 195445 (goo notes). In .ctb the slicer-info block carries a capability byte, 0x40 for version 4 and 0x50 for version 5 (Chitu notes).

No file says whether the machine on the other end obeys it. Chitu firmware reads the tables from 4.3.9 on (Chitu notes). The profile default is “reads them”. The profile generator clears it for the 26 machines whose published profile asked for a .ctb older than version 4, and points them at version 4 instead (profiles, “Where the numbers come from”). Version 4 needs firmware from 2021 or later; an older board refuses the file rather than misprinting it (Chitu notes).

With the flag off, Encrust never sets the advance flag, and refuses a job that needs the tables, such as an exposure band by height, rather than write it at header exposure silently. The error says does not read the per-layer tables, so exposure cannot vary by height. A bottom ramp alone still works: the machine ramps the bottom block itself from the header’s bottom exposure, normal exposure and transition count.

The switch is Per-layer settings, on the printer’s form under Settings, Printers.

Variable layer height

A stack of layers of different thickness (adaptive layers) only prints right if the plate moves to each layer’s own Z. A machine that steps by the header’s layer height instead can’t print the stack.

Encrust’s profile default is off, and no shipped profile turns it on: it is, in the notes’ words, “the user’s own tested claim” (profiles, “What the firmware obeys”). The notes give the reason: machines of one maker and format differ. A job with mixed thicknesses is refused without the claim, in the window and on the command line, where the error names --adaptive. The switch is Variable layer height. The notes give no firmware version for it.

The vat that tilts

The Saturn 4 Ultra is the profile where the flag was found by printing, and its 16K sibling and the Mars 5 Ultra carry the same setting. The plate stopped rising at around 0.6 mm, just past the bottom and transition layers, while the panel kept exposing. The same job with per_layer_settings = false printed to the end. The two files differed in one meaningful byte, offset 195445, the advance mode. Layer count, Z per layer, exposure ramp, lift and retract were identical.

The reading, per our notes: a tilting vat does not pull the plate off the film, so the 5 mm lift and 5 mm retract the per-layer records carried, inherited from peel machines, are not a motion it can perform, and advance mode is what makes the firmware obey them (profiles, “What the firmware obeys”). The evidence points at the flag, not at the motion values: we did not test zeroing the lift instead.

The cost is real. This profile now refuses exposure bands and adaptive stacks outright, because the machine reads only the header. The vendor’s file for the machine clears the same byte (goo notes).

[firmware]
per_layer_settings = false

What the notes say about firmware versions

Two versions, and no more: Chitu firmware reads per-layer tables from 4.3.9, and a version 4 .ctb wants firmware from 2021 or later. The .goo container takes only V3.0 (goo notes). For the Saturn 4 Ultra, variable layer height, the machine-name match and the .cxdlp model code, the notes record behaviour and no version. Don’t read an absent version as “all of them”.

Counted from the profile files, not measured on 199 machines: 167 read per-layer tables (163 say so, four have no firmware block and take the default), 32 do not, and none claims variable layer height.

Sources