Hi guys ,
This the ChatGPT answer
- Measured minimum ≠ true optical minimum
The software measures an approximation of sharpness (usually HFR – Half Flux Radius).
However:
The measurement is noisy (seeing, turbulence, sensor noise)
The stars used can vary from one frame to another
Sky background and guiding also affect the measurement
👉 Result: the lowest measured point can be slightly offset from the true focus.
📉 2. Mathematical curve fitting
ASIAIR doesn’t simply pick the lowest measured value.
It performs a mathematical fit (often a parabola or V-curve) to estimate:
👉 the theoretical minimum of the curve
This minimum can:
lie between measured points
be influenced by the overall shape rather than a single value
👉 So the final focus point may differ from the lowest raw measurement.
🌫️ 3. Seeing (atmospheric turbulence)
Even with a perfect setup:
turbulence randomly “bloats” stars
some frames appear artificially better or worse
👉 A point can look best just by chance.
So the software prefers a global trend rather than a single minimum.
⚙️ 4. Mechanical backlash and hysteresis
With a focuser like the ZWO EAF:
there can be mechanical play (backlash)
the actual position depends on the direction of approach
👉 ASIAIR typically applies a strategy such as:
always finishing focus from the same direction
compensating for backlash
So the final position may be slightly offset to ensure mechanical stability.
🌡️ 5. Drift during the autofocus run
During autofocus:
temperature may change
focus can drift slightly
the optical tube can expand or contract
👉 The true minimum is actually moving during the process.
🧠 Summary
The final focus point chosen by ASIAIR is:
👉 a robust estimate of best focus,
not simply the lowest measured value.
It balances:
measurement noise
mathematical modeling
mechanical stability
real-world conditions
✅ Practical advice
Don’t try to force focus to match the lowest point
Instead, check for:
a clean, symmetrical curve
a well-defined V-shape
If the curve is messy → something is off (seeing, saturated stars, too few samples)
