Dithering

Dithering

1️⃣ What is dithering?

Dithering is the deliberate, small movement of the telescope or lens between groups of exposures, shifting the image by a few pixels each time.

The stars stay aligned when stacked — but fixed sensor defects do not.

This is the single most important technique for removing:

• Banding
  
  

• Walking noise
  
  

• Hot pixel streaks
  
  

• Residual amp glow
  
  

• Pattern noise that does not average out naturally
  
  

2️⃣ What dithering actually does (and why it matters)

Most noise behaves well:

• Shot noise averages down with √N

• Read noise becomes irrelevant once sky noise dominates

But pattern noise does not average down.

Without dithering:

• Hot pixels walk across the frame

• Banding aligns between subs

• Rejection algorithms fail because defects appear “real”

With dithering:

• Each defect lands in a different place

• Sigma-clipping / Winsorized rejection works properly

• Pattern noise collapses rapidly

Dithering turns un-averageable noise into averageable noise.

This is especially critical for:

• DSLRs

• Uncooled OSC cameras

• Long focal lengths

• Warm nights

3️⃣ How to dither (guided and unguided)

✔️ Guided dithering (EQ mounts)

This is the simplest and most precise method.

You need:

• A guide camera

• PHD2, ASIAIR, or similar

• Mount control (ST-4 or pulse guiding)

How it works:

• Guiding software issues a random offset

• Mount moves a few pixels

• Guiding resumes and recenters

• Imaging continues automatically

Recommended setup:

• Dither every 10–15 minutes, not every sub

• Set settle time so guiding stabilizes before the next exposure

✔️ Unguided dithering (Star Adventurer, trackers)

You can and should dither unguided.

There are three practical methods:

A) App-based dithering

• Star Adventurer Wi-Fi / Snap port modes

• Use small random RA offsets

• Add a settle delay

B) Intervalometer dithering

• Insert a pause between subs

• Manually jog RA slightly

• Resume imaging

C) Manual dithering

• Physically nudge RA

• Or re-frame slightly

• Resume imaging

This works because:

• You only need ~16 total dithers

• Exact randomness is not critical

• Median stacking + rejection does the rest

4️⃣ How much dithering? (the rule)

≈ 10 pixels per dither

This is:

• Independent of focal length

• Independent of pixel scale

• Large enough to break pattern correlation

• Small enough to avoid framing loss

Think in pixels, not arcseconds.

Smaller dithers (<5 px) are often ineffective
Larger dithers (>20 px) waste field of view

5️⃣ Settle time matters (especially unguided)

After a dither:

• RA backlash must recover

• Periodic error must stabilize

• Tracking must “catch up”

Always include a delay between shots:

• Guided: use guider settle time

• Unguided: 5–10 seconds minimum

• Long focal lengths: err on the longer side

Skipping this causes:

• Star elongation

• Inconsistent sub quality

• Worse rejection performance

6️⃣ Manual dithering strategy (simple & effective)

You do not need to dither every sub.

Noise reduction has diminishing returns.

A practical rule:

• ~16 dithers total per session

• After that, gains are minimal

Example:

• 4-hour session

• Dither once every 15 minutes

• That’s enough

Also count:

• Meridian flips → automatic large dither

• New sessions → fresh alignment = fresh dither

Median and sigma-clipped stacks love this.

DSSq Bottom Line

• Dithering is mandatory, not optional

• It matters more than dark frames for uncooled cameras

• 10 pixels is the sweet spot

• You only need ~16 total dithers

• Unguided users are not excluded

If you don’t dither, you are fighting noise that cannot be averaged away.