A white halo is a symptom, not a setting. Dusty pixels usually start in dirty alpha; a clean even border can come from an underbase mask extending beyond colour; a one-sided duplicate edge points toward colour-to-white registration or scaling; and a soft design can expose a threshold decision. Identify the geometry in artwork, white preview and RIP before touching choke.
Read the halo’s shape before changing a value
Put the same suspect edge under magnification in four places: the approved colour artwork, its alpha/mask, the generated or supplied white plane, and the target RIP preview. The pattern across those views is more useful than a copied “best choke” number.
| Visible symptom | Likely first branch | What confirms it | Why choke alone is incomplete |
|---|---|---|---|
| Dust, speckles or a faint uneven box around the design | Low-opacity alpha debris or hidden background-removal residue | Flecks appear in the alpha/diagnostic view and often enter alpha-derived white | Choke erodes eligible nearby mask edges; it does not decide that distant RGB/alpha fragments are unwanted artwork |
| Thin, clean and broadly even white outline around hard colour edges | White-mask footprint extends farther than visible colour, or downstream mapping changes the footprint | Colour and white overlays share an origin, but the white boundary is consistently outside | Choke may be the relevant geometry control, but scale, spread/trap and RIP-side generation still need checking |
| White visible mainly on one side, with the opposite side tight or colour overhanging | Mechanical colour/white registration, transport, origin or unequal scaling | The offset has a direction and may repeat consistently across unrelated artwork | Symmetric erosion cannot realign two planes or correct printer mechanics |
| Soft glow, shadow, smoke or feather changes character at the edge | Intentional semi-transparent artwork meets a threshold/underbase rule | The approved alpha contains a continuous fade rather than isolated contamination | More choke can amputate the fade; threshold and white-density behavior need an artistic decision |
Do not diagnose from the pressed garment alone. Fabric, cure and press can change the appearance, but the file and RIP views can tell you whether the geometric defect already existed before output.
The transparent-PNG alpha guide covers the first row in depth. For broader intake size, colour and transparency checks, use the DTF file requirements.
Diagnose in a fixed order
1. Inspect colour and alpha at the source
View the production candidate over black, white and a saturated background. Look beyond the intended edge for isolated pixels, a faint rectangle, coloured matte residue and excessive transparent padding. Switch to an alpha or mask view if available.
Classify soft pixels before deleting them. A disconnected low-opacity speck far from the design is different from a continuous shadow or anti-aliased curve. Automatic removal can clean the first and damage the second.
2. Inspect the white plane by itself
Turn off the colour composite and study the underbase silhouette. Ask:
- Does white follow only intended artwork, or also background debris?
- Is the boundary evenly outside, evenly inside or offset in one direction?
- Do thin strokes retain enough underbase?
- Does an intentional fade become an abrupt hard plate?
- Are the canvas dimensions, crop and mirror state identical to colour?
The white-channel preview guide explains what to look for without relying on the colour thumbnail.
3. Overlay colour and white
An overlay distinguishes symmetric geometry from directional displacement. If white is wider by roughly the same amount on every hard edge, investigate underbase generation, spread/trap and choke. If white escapes on the right but colour escapes on the left, investigate alignment or scaling. If only scattered pixels light up, return to source alpha.
4. Recheck the target RIP import
Confirm physical width/height, origin, mirror state, crop, plane names, polarity and ink mapping in the exact queue. “Fit to media” or unequal interpretation can move or rescale a correct source separation. A prepress preview cannot prove that a later import preserved the geometry.
5. Isolate the output system with a controlled test
Use a small test target with hard rectangles, thin strokes and a known soft transition. If the RIP preview is aligned but the printed colour and white separate directionally, investigate the printer/transport/queue calibration rather than increasing a file choke until the error is hidden.
Four controls that solve different problems
In an alpha-derived white workflow, the order matters. NestSheet’s current generation path is alpha → threshold → base mask → choke → mode/halftone → amount. A later control cannot reconstruct information removed earlier.
| Control | What it changes | What it cannot fix |
|---|---|---|
| Alpha threshold | Which source-opacity values enter the initial white mask | It does not move a white plane, align heads, remove opaque RGB backgrounds or improve source resolution |
| Choke | Erodes eligible generated-underbase geometry inward by a physical amount implied by pixels and PPI; the current NestSheet engine protects near-white source-art pixels from erosion | It does not remove distant debris, change colour artwork, repair directional registration or guarantee thin-stroke coverage |
| White amount | Scales white coverage/density within the selected mask | It does not move the mask boundary or align colour and white |
| Mechanical / RIP registration | Aligns colour and white origins, scale and output timing through the relevant device/queue controls | It does not clean alpha or decide the intended shape of a soft underbase |
Threshold is sometimes mistaken for choke because both can reduce visible white. The diagnostic difference is topology. Raising a threshold can make disconnected low-alpha regions disappear and can change fades throughout the artwork. Choke moves eligible generated-underbase boundaries inward. In the current NestSheet engine, near-white source-art pixels are protected because there may be no colour ink to hide behind; narrow non-white-backed features can still disappear even when their source alpha is solid.
White amount is sometimes mistaken for geometry because a lighter halo is less visible. The footprint has not changed. If the mask extends beyond colour, density reduction merely makes the same shape fainter and can weaken the intended underbase everywhere else.
Convert pixel choke to a physical distance
A pixel setting has no portable physical meaning without the working image resolution:
millimetres per pixel = 25.4 ÷ PPI
physical choke in mm = choke pixels × 25.4 ÷ PPIAt 300 PPI, one pixel is approximately 0.0847 mm and two pixels are approximately 0.169 mm. At 150 PPI, the same two pixels represent approximately 0.339 mm. This is why copying a pixel value from a workflow using another raster density changes the physical erosion.
PPI here describes image pixels at output size; it is not printer device DPI. Also remember that a scaled RIP import changes the physical meaning of the already-generated pixels. Verify final dimensions before comparing tests.
NestSheet’s current effective choke engine range is 0–10 px and the current DTF product default is 2 px. One current Prep control extends visually to 20 px, but the live preview contract rejects values above 10 and the output generator clamps at 10; until those surfaces align, treat values above 10 as unsupported. These are product facts, not universal recommendations. Select a value from the measured defect and a controlled test, with special attention to fine strokes and type that erosion can remove.
Use Prep and export preview for the right proof
In the current NestSheet Prep workflow, ART, WHITE, FABRIC and DIFF views let the operator compare source colour, white and visible consequences. For supported tuning values, including choke from 0–10 px, alpha-derived preview and export resolve the same white tuning chain, so threshold, choke, mode and amount should be judged together before packing the full batch.
Eligible file-supplied TIFF white planes follow a different boundary: they are supplied production data rather than a newly inferred alpha mask. Inspect those planes and the downstream RIP mapping directly; do not assume an alpha-control preview is byte-for-byte proof of arbitrary supplied channels.
The product-level white-underbase workflow shows where the controls sit. The safe release sequence is source alpha → white-only view → colour/white overlay → export dimensions and planes → target RIP preview → controlled film/press test.
Sources
- Caldera HelpDesk, Choke settings and color management in DTF printing: https://helpdesk.caldera.com/hc/en-us/articles/29224655665553—Choke-Settings-and-Color-Management-in-DTF-Printing
- Caldera HelpDesk, Create a white underbase for DTF and DTG applications: https://helpdesk.caldera.com/hc/en-us/articles/25408381491217-How-to-create-a-white-underbase-for-DTF-and-DTG-applications
- Caldera HelpDesk, Create a white underlay in DTF: https://helpdesk.caldera.com/hc/en-us/articles/15207936811921—How-to-create-a-white-underlay-in-DTF
Caldera references were captured for the broad underbase/choke concepts in July 2026; their pages blocked direct automated retrieval, so no vendor-specific setting, quotation or universal range is inferred here. NestSheet defaults and behavior were verified against shipped code on 21 July 2026.
