# Quality issues: jaggy edges, halos, color shifts, and other visual problems

URL: https://helpcenter.readypixl.com/articles/quality-issues
Markdown: https://helpcenter.readypixl.com/articles/quality-issues.md
Section: Troubleshooting
Summary: When the pipeline runs successfully but the output looks wrong — visible halos, jagged edges, color shifts, blurry results. Diagnosis and fix per problem.

Sometimes the pipeline runs without errors but the output **doesn't look right.** Halos around your subject. Edges that look chopped. Colors slightly off. Images that look blurry.

This article diagnoses the common visual quality issues + how to fix each.

## Visible halo around the subject

**What it looks like:** a faint ring of color or gray pixels around the cut-out subject. Most obvious on dark backgrounds (in your final use) where it shows up as a glow.

**What's happening:** Color Removal V2 didn't fully erase the background pixels closest to your subject's edge. Those leftover semi-transparent pixels show as a halo.

**Fix in this order:**

1. Add **Transparency Cleaner** after Color Removal V2 and set its slider to **30-50**. Targets exactly this issue.
2. Open Color Removal V2's **Advanced** toggle and add some **Halo Width** — shaves the edge pixels off.
3. Bump Color Removal V2 to a higher **Preset** (or raise Base Threshold a few points — presets are Low 6, Medium 8, High 10). The halo is partly because the threshold was too low.
4. For really stubborn halos, also add **Speckle Remover** with default Max cluster size 50.

## Jagged / stepped edges (instead of smooth)

**What it looks like:** the cut-out subject has visibly stair-stepped edges, like 1990s pixel art.

**What's happening:** Color Removal V2 made hard yes/no decisions per pixel — every pixel is either fully kept or fully erased. No softness in between.

**Fix:**

- Color-based removal makes hard per-pixel cuts, so for smooth edges on hair, fur, or soft subjects use **AI Background Remover** (Recraft model, 15 credits).
- Or try a gentler **Preset** (Low) and pick the background color in **Manual** mode for a cleaner cut.
- Run **Transparency Cleaner** afterward to tidy any leftover rough edge pixels.

## Colors look slightly different after the pipeline

**What it looks like:** brand colors that shifted, neutrals that look warm/cool.

**What's happening:** Image Adjustment shifted them, or your output format compressed them differently than the source.

**Fix:**

- Check if you have an Image Adjustment step. Try removing it temporarily — does the color look right? Then your Adjustment was the cause.
- If outputting to JPEG, JPEG compresses some color subtly. Switch output format to PNG for color-critical work (logos, brand-color designs).
- For consistent output across many images with different source quality, use Image Adjustment to normalize, accepting the small shift as a tradeoff.

## Image looks blurry / soft

**What it looks like:** less crisp than the source. Edges are softer than expected.

**What's happening:** several possible causes:

1. **You upscaled.** Resize beyond 100% adds pixels that didn't exist — necessarily soft.
2. **You applied Blur** filter (intentionally or by accident).
3. **JPEG compression.** Saving as JPEG with low quality compresses detail. Use PNG for crispness.
4. **Source was already soft.** ReadyPixl can't add crispness that wasn't in the source.

**Fix:**

- For #1: don't upscale with Resize — use AI Upscale, or design at the target size from the start.
- For #2: remove the Blur step.
- For #3: use PNG output, or if JPEG is required, use higher JPEG quality.
- For #4: add **Filters → Sharpen** at strength **20-40** to recover some crispness.

## Image has a color cast (everything tinted yellow / blue / pink)

**What it looks like:** the whole image has a wash of one color — usually yellow (warm), blue (cool), or pink/green (off-white-balance source).

**What's happening:** the source image was generated or captured under wrong white-balance assumptions.

**Fix:** add Image Adjustment with channel corrections:

| Cast | Settings |
| --- | --- |
| Yellow / warm | Blue +10, Red -5 |
| Blue / cool | Red +10, Blue -10 |
| Pink / magenta | Green +10, Red -5 |
| Green | Red +5, Blue +5, Green -10 |

Tune from the suggested starting point until the cast is gone.

## Output is way too dark or too bright

**What it looks like:** dark areas are crushed black or bright areas are blown out white.

**What's happening:** Image Adjustment Brightness / Contrast got over-applied, OR your source image was extreme.

**Fix:**

- Check Brightness and Contrast — defaults are 0; large values shift the image significantly.
- Bring both back to 0 and re-tune in smaller increments (±5 to ±15).
- For genuinely dark sources, raise Brightness +20 and Contrast +5 together.

## Subject is off-center after Reposition

**What it looks like:** subject is sitting in the top-left corner or other unexpected location of the canvas.

**What's happening:** Reposition's Alignment got changed from Center, the container was moved, or stray see-through pixels are throwing off the placement.

**Fix:** set Reposition's **Alignment** to **Center** and add **Trim** before Reposition to remove stray pixels. Reset it unless you specifically want a corner placement.

## Subject got cropped off the canvas

**What it looks like:** part of your subject is missing where it should hit the canvas edge.

**What's happening:** subject is bigger than the canvas you're placing it on.

**Fix:**

- Add a **Resize** step before Reposition to scale your subject down first.
- Or increase the Reposition canvas size.
- Or lower Reposition Padding (less border = more room for the subject).

## Random small holes / dots in the design

**What it looks like:** tiny see-through holes scattered through your design where there shouldn't be.

**What's happening:** Color Removal V2 erased pixels inside your design that happened to match the background color. Or Speckle Remover ate small intentional details.

**Fix:**

- Turn ON Color Removal V2 **Contiguous** — only removes background pixels connected to the edge.
- Lower Speckle Remover **Max cluster size** to 25 or below — won't eat small intentional details.

## Watermark text is unreadable

**What it looks like:** watermark blends into the design, or is too small to read.

**What's happening:** color and size mismatch with the underlying image.

**Fix:**

- Pick a watermark color that contrasts with most of your designs. White on dark, black on light, or semi-transparent gray for mixed batches.
- Bump font size to 32-50 px for medium images, 60-80 for big images.
- Increase Opacity to 70+ if subtlety isn't the goal.

## Distress / Frames look weird on this batch

**What it looks like:** the texture or frame doesn't match the proportions of your design — looks stretched or oddly placed.

**What's happening:** the mask file is being scaled to fit a different aspect ratio than it was designed for.

**Fix:** use a mask designed for your design's aspect ratio. Square masks for square designs, wide masks for wide designs.

## When the issue is your source image

Sometimes the pipeline is working correctly and the source image is the problem:

- **Low resolution source** → low-resolution result. Pipeline can't add detail.
- **Bad lighting source** → bad lighting result. Image Adjustment can correct *some* but not transform poor lighting into good.
- **Out of focus source** → out of focus result. Sharpen helps a little but can't recover a truly blurry source.

For these, **fix the source and run again** — it's faster than trying to fix in the pipeline.

## What to read next

- **[Per-tool quirks](per-tool-quirks)** — when the issue is a specific tool acting up
- **[Common error messages](common-errors)** — for actual error toasts
- **[Performance tips](performance-tips)** — for slow / memory issues