A useful DTF test file is a small, declared target that can fail in recognisable ways. Give it known physical dimensions, hard and soft-alpha edges, disclosed fine strokes and type, expected-no-white and colour-backed expected-white regions. Compare the same target from source through Prep, the actual export and exact RIP preview, then print it. A clean screen preview is necessary evidence, never a physical guarantee.
Define the decision before drawing the target
This is a go/no-go protocol for one file and handoff, not a broad upload checklist or printer-calibration procedure. Use the DTF file-requirements guide for source, resolution, colour, and container questions.
Before creating samples, write the decision in one sentence: “Release this exact source, Prep recipe, output row, RIP/version/queue, and physical stack only if every declared target feature passes its recorded check.” Then define “passes” from the shop’s existing controlled standard. This article supplies no universal target size, type size, stroke width, alpha threshold, choke, white amount, scale tolerance, or registration tolerance.
Assumptions must travel with the result:
- physical width and height are recorded in millimetres or inches;
- source pixel dimensions, effective PPI at that size, and any embedded density are recorded separately from printer DPI;
- the NestSheet working comparison uses 8-bit RGBA artwork, while the final output container, bit depth, composite, alpha, and production planes are recorded as actually emitted;
- white source is declared as alpha-derived or file-supplied, with names and polarity recorded when separate planes exist;
- the exact RIP product, version, queue, scaling state, and device resolution/mode are named; and
- printer, ink, film, powder, cure, press, substrate, test size, and observation method are fixed for physical comparison.
No downloadable calibrated target is included here. Build and version the target used by the shop, keep its checksum with the record, and revise it only through a new comparison.
Put one diagnostic job in every region
Do not fill the target with attractive sample art. Each region should answer one question and have an expected digital state before any print is made.
| Target region | Declare in the source | What to compare | What it cannot prove alone |
|---|---|---|---|
| Dimension-marked hard shape | Pixel size, intended physical width/height, fully opaque edge | Source size → RIP-reported size → measured output | Fine-detail survival or colour accuracy |
| Intentional soft-alpha transition | Start/end opacity and approved visual shape | Source alpha → Prep alpha/white → export → RIP preview | Whether film, powder, and press will reproduce the fade |
| Transparent no-art space | Expected alpha zero and no production ink | Alpha and plane previews at that location | That every application interprets all transparency identically |
| Fine strokes and type | Actual source widths and final physical sizes, without calling them universal minima | Full-resolution raster, imported preview, physical sample | A general safe tolerance for other devices or art styles |
| Expected no-white region | Why white should be absent | White view or exported plane, then queue mapping | Correct physical output without a print |
| Colour-backed expected-white region | Colour shape and intended supporting white footprint | ART/WHITE comparison, exported representation, RIP preview, print | A universal choke or density setting |
| True white-only region | A separate spot-bearing fixture with declared plane semantics | Exported plane and exact queue mapping before print | Acceptance by an untested RIP or container variant |
Put fine-detail candidates in a disclosed series around the shop’s own decision boundary. Label source and intended physical widths; do not label an unprinted value “safe.”
The true white-only region needs its own spot-bearing fixture because ordinary RGBA transparency and a named white production plane are different signals. The spot-channel guide owns the detailed naming, polarity, and mapping model. The target only records whether the expected region survives this exact handoff.
Establish scale without treating metadata as output
The basic arithmetic is:
effective PPI = source pixels ÷ intended inches, where inches = millimetres ÷ 25.4.
A 1,200-pixel-wide shape intended to measure 101.6 mm is 1,200 ÷ 4 = 300 effective PPI. That is a calculation example, not a 300-PPI acceptance rule. Changing a density field without resampling changes the intended physical interpretation; resampling changes the pixel count. Neither guarantees that the RIP will honour the metadata.
PNG can carry intended physical pixel size in its pHYs chunk, but that information may be absent or use an unknown unit. Compare queue-reported dimensions and measured output rather than trusting one metadata field.
Record both axes. If the intended aspect ratio changes, or the RIP reports a different physical size, stop before judging strokes, alpha, or white. A scale error changes the physical meaning of every pixel-based feature downstream.
Checkpoint 1: approve source colour and alpha
Open the versioned source over black, white, and a saturated temporary background. Confirm the hard edge is hard, the soft transition is intentional, holes are transparent, and distant no-art space contains no unexpected opacity. Adobe documents that anti-aliasing partly fills edge pixels and that former-background colour can remain in those fringe pixels; that explains the diagnostic, not whether a particular edge will print successfully.
Use the transparent-PNG alpha guide for the contrast and mask inspection. Do not clean all partial alpha merely because it is partial: the soft sample exists specifically to prove how the workflow treats intended opacity.
At this checkpoint, save the source checksum, pixel dimensions, physical intent, colour/alpha description, and the width of every fine-detail sample. If the source does not match its own declaration, there is no reason to continue downstream.
Checkpoint 2: compare source with Prep
Open the same file in Prep without changing its intended physical size. Use ART for the working colour, WHITE for the current white review, and the ALPHA overlay for opacity. SOLID, FRINGE, FADE, TRASH, and DEAD overlays locate classes of pixels; they do not decide whether a pixel belongs to the approved design.
Inspect hard edges, the soft transition, no-art space, fine detail, and both white-expectation regions. The print loupe replays enabled Prep steps at full resolution, displaying one image pixel as one CSS pixel; use it instead of the downscaled ordinary preview.
Prep’s THIN scan accepts an operator-entered millimetre value and marks detected zones. Enter the shop’s declared test boundary; do not copy internal defaults or risk labels as press-validated tolerances. The result is a locator for review, not an automatic go/no-go score.
For alpha-derived white, the current generation order is alpha → threshold/mask → choke → mode/halftone → amount. Compare the ART and WHITE views at every hard, soft, fine, expected-white, and expected-no-white region. The white-underbase page describes the current controls, while the halo diagnostic explains why dirty alpha, mask geometry, and physical displacement remain separate branches.
File-supplied white has a stricter proof boundary. Prep can show a combined review film for eligible supplied white, but that view is not the serialized per-plane export bytes. Inspect the actual emitted planes separately. Do not infer that colour and spot data share the intended relative geometry merely because the Prep view looks correct.
Record every enabled Prep step and parameter. If an edge changes here, the failure is already upstream of export and the RIP. Resolve or approve that change before proceeding.
Checkpoint 3: inspect the emitted production artifact
Export through the exact row intended for production, not a convenient proof format. The current export matrix separates colour-plus-alpha output from spot-bearing TIFF or flattened PSD/PSB rows; those structures are not interchangeable.
Reopen the emitted file in an independent reader that can expose the representation being tested:
- Record the export checksum, pixel dimensions, declared physical dimensions, container, bit depth, colour mode, and channel count.
- For colour-plus-alpha output, compare the composite and alpha with the source and Prep checkpoints.
- Where the container and reader expose separate production planes, record their names, polarity, dimensions, and observed position relative to colour.
- Check every expected-white and expected-no-white region directly in the emitted data.
- Confirm the hard/soft edges and fine-detail candidates still exist at the expected pixels.
An export-size guard proves only that NestSheet accepted the requested sheet/container combination. It does not prove downstream import or printing. A reader displaying colour does not prove that it found every production plane.
If the artifact cannot be inspected at the level required by the handoff, the result is not proven, not “pass.” Choose an inspectable path or obtain a controlled import receipt before full production.
Checkpoint 4: import into the exact RIP version and queue
Use the actual production RIP, version, queue, and profile. The RIP hub explains the prepress-to-RIP boundary, but no general page can certify an untested installed configuration.
Start from a documented actual-size or no-scaling state when the queue provides one; otherwise record every size or fit transformation the queue reports. Then inspect:
- reported physical width and height;
- the colour composite once, without an unexpected production mask flattened into it;
- alpha or each intended production plane, where that workflow exposes them;
- plane names, polarity, and mapping to the intended ink role;
- expected-white and expected-no-white regions;
- relative colour/white geometry as an observation to be tested, not an assurance; and
- hard edges, the soft-alpha transition, and fine-detail candidates at the closest available preview scale.
The white-channel preview guide is a visual-reading aid. Its screen view still cannot prove physical output, and any unsourced universal pixel or density prescription is outside this protocol.
Use the first divergence to choose the next branch:
| First checkpoint that differs from the declaration | Likely investigation branch | Release action |
|---|---|---|
| Source | Target construction or approval | Stop; correct and re-version the source |
| Prep | Enabled edit, alpha interpretation, or white-generation choice | Stop; isolate one Prep parameter and rerun |
| Exported artifact | Output row, container, alpha/plane serialization, or scaling | Stop; inspect a newly emitted artifact |
| RIP preview | Import scale, plane discovery, polarity, mapping, or queue processing | Stop; verify the exact queue/manual and controlled import |
| Physical output only | Printer/transport, ink, film, powder, cure, press, or substrate | Hold the file constant and investigate the physical path |
This table narrows a fault; it does not prove one cause from one observation. Repeat the sample when measurement uncertainty could change the branch.
Checkpoint 5: print the same small target
Print the artifact already inspected—do not regenerate it between preview and test. First compare the film with the declared target. If transfer quality is part of the decision, press one sample with the recorded substrate and process, then inspect it using the shop’s declared method.
Measure the known-dimension shape in both axes. Check the hard edge, intentional soft transition, fine-detail series, no-white region, colour-backed white region, and separately declared white-only fixture. Record observations, not adjectives: measured size, which labelled strokes/type remain distinguishable, which alpha steps remain distinct, and what white appears in each expected region.
The go/no-go record should be explicit:
| Gate | Go only when | No-go response |
|---|---|---|
| Scale | RIP report and measured output meet the shop’s predeclared tolerance | Fix the first scaling/resampling divergence and retest |
| Alpha | Hard and soft samples match the approved intent at the required checkpoints | Isolate source cleanup, Prep, export, or RIP interpretation |
| White | Expected regions are present/absent with the declared mapping and physical result | Recheck source, generation, emitted data, mapping, then device path |
| Fine detail | Predetermined labelled candidates survive to the shop’s criterion | Keep the file fixed and bracket one responsible variable |
| Physical process | Film and pressed sample, where required, meet the declared process checks | Do not release a full sheet from a screen-only pass |
A clean source, Prep view, export, and RIP preview cannot guarantee ink delivery, colour-to-white physical registration, film coating, powder behavior, curing, pressing, substrate response, or wash performance. Those are controlled-print uncertainties. When one variable changes—file, recipe, output row, RIP version/queue, printer mode, material, or process—rerun the smallest relevant part of the protocol before release.
Sources
- W3C, PNG Specification, Third Edition: https://www.w3.org/TR/png-3/
- Adobe Photoshop, Printed image resolution: https://helpx.adobe.com/photoshop/desktop/crop-resize-transform/resize-adjust-resolution/printed-image-resolution.html
- Adobe Photoshop, Fringe pixels around a selection: https://helpx.adobe.com/photoshop/desktop/make-selections/refine-modify-selections/fringe-pixels-around-a-selection.html
- Adobe Illustrator, Print color separations: https://helpx.adobe.com/illustrator/using/printing-color-separations.html
- CADlink Digital Factory, Transparency Opacity Tool: https://help.cadlink.com/website/digital_factory/en/bonus_documentation/transparency_opacity_tool.htm
- CADlink Digital Factory, Queue menu — Output: https://help.cadlink.com/website/digital_factory/en/production/menus/queue_menu_output.htm
External specifications and vendor documentation were checked July 2026. They support alpha/size semantics, preview limits, and a test-first boundary—not universal DTF tolerances. NestSheet’s Prep views, diagnostic overlays, full-resolution loupe, white-generation order, and export-guard boundary were checked against shipped code on 21 July 2026. No named-RIP import or complete physical-print receipt was available.
