Photons to photos noise chart - am i reading this right?

Started 3 months ago | Discussions thread
mostlyboringphotog Forum Pro • Posts: 10,057
Re: Photons to photos noise chart - am i reading this right?

swimswithtrout wrote:

Tristimulus wrote:

mostlyboringphotog wrote:

swimswithtrout wrote:

Tristimulus wrote:

Would ignore the charts completely for astrophotography.

For older cameras without dual gain ISO1600-3200 usually works best for deep sky. Simply find what works for you. This is fine for cameras like my astromodded Canon 6D.

For dual gain image sensors the second gain step works fine for astrophotography as this is where read noise reach bottom level. This works wonders for my Sony cameras.

Then there's ISO- invariant cameras with a flat line graph. No need to ever shoot over unity gain which is usually not base ISO. - bold added

I'm afraid I disagree, especially shooting DSO. Much of the image information is in the low end values of the ADC data.

Let's say it is between 8 and 511, 504 values gradation at ISO 1600. Now, shoot at ISO 400, at the same exposure which will result in values from 1 to 63, you will get only 64 values of gradation. The ISO 400 image can be brightened in PP by 3 stops to match the brightness of ISO1600 image but the different values in the ISO400 image is still 64.

Well - raising ISO does NOT enhance the sensitivity of the image sensor.

What is needed to collect more light witha given setup is a longer exposure.

That is, value of 1 in ISO400 is increased to 8 but the value of 2 is increased to 16 and the image has no values between 8 and 16, whereas the ISO1600 does have the values between 8 and 16.

The above may show up as "crunchy" or even banding like using 65K colors instead of 16M colors.

My camera is "ISO-less" and has a unity gain of ISO 320, I shoot either ISO 200 or 400. At unity gain, one photon in equals one electron out. Increasing the ISO only amplifies that signal. You gain no more new information. In fact, you lose information raising the ISO in the form of deceased dynamic range....

If the signal is not conditioned to the full scale input range of the ADC, you would lose more than if you did.

Increasing the ISO loses DR only if the scene is bright enough to clip the highlight - in that case your ISO setting is too high.

And ISO-less is really that the read noise is low enough so that higher ISO does not improve the read noise.

However, ADC still requires the full scale input to realize the maximum DR and SNR and one way to do that is to amplify the input signal which is usually done by increasing the ISO dial.

This has nothing to do with the ISO sensitivity. It's about the ADC.

"For a sine wave, we can find the maximum SNR of an ideal N-bit quantizer as SNR=1.76+6.02N."  - bold added

If the camera has 14bit ADC and if the input range reaches the full scale value (14bit), then 86dB can be realized.

If by lowering the ISO and thus lowering the analog gain so that input to the ADC reaches only the 10bit, then only 62dB of SNR is realized due to the increased quantization noise.

It's not about gaining more SNR; it's about losing the least SNR.

In most cases, signal processing loses the SNR and DR due to noise; the goal is to lose the least amount.

 mostlyboringphotog's gear list:mostlyboringphotog's gear list
Panasonic Lumix DMC-GM1 Pentax 645Z Nikon 1 J5 Fujifilm GFX 50R Canon EOS RP
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow