Test quattro overexposed sky.

Tests recovery
details in HL from a 400 ISO RAW.

Original photo made by:

http://www.photographyblog.com/previews/sigma_dp3_quattro_photos/

Original photo:

In the shadows:

http://www.dpreview.com/forums/post/56129125

http://www.dpreview.com/forums/post/56163332

Wow! That isn't supposed to be possible, from what we have been led to believe. I thought it was the shadows that were supposed to be recoverable, when shooting with a Quattro. This looks like one-shot HDR! What's the dynamic range of the new sensor again?

Very cool!

I was able to recover about 1 stop overexposure from a DP0Q shot.  How did you go about recovering this overexposure?

I'm a bit confused, what is the difference between the three conversions?  Are they from different software?  What settings were used to recover HL?

Tried it myself - these settings recovered all the overexposed bit and gave a flat contrast TIF to then work on:

Exposure: -1.0

Contrast: -1.5 (ex- 0.0)

Shadow: +0.5

Highlight: +0.5

Saturation: 0.0

Sharpness: 0.0

X3 Fill Light: +0.2

While I don't have a Quattro (yet) I've now tried several X3F's and have been perplexed by much of the criticism I've seen posted around forums about poor highlight recovery etc.

maceoQ wrote:

digi2ap wrote:

Tried it myself - these settings recovered all the overexposed bit and gave a flat contrast TIF to then work on:

Exposure: -1.0

Contrast: -1.5 (ex- 0.0)

Shadow: +0.5

Highlight: +0.5

Saturation: 0.0

Sharpness: 0.0

X3 Fill Light: +0.2

with or without overexposure correction?
(i don't think you can recover the clouds without)

Without.

maceoQ wrote:

digi2ap wrote:

maceoQ wrote:

digi2ap wrote:

Tried it myself - these settings recovered all the overexposed bit and gave a flat contrast TIF to then work on:

Exposure: -1.0

Contrast: -1.5 (ex- 0.0)

Shadow: +0.5

Highlight: +0.5

Saturation: 0.0

Sharpness: 0.0

X3 Fill Light: +0.2

with or without overexposure correction?
(i don't think you can recover the clouds without)

Without.

Strange. I get the result on the left with your settings. (right side is a Kalpanika conversion)

left side, SPP without highlight correction; right side Kalpanika

Agreed there is something very different in these two conversions!!

The over exposure correction is making no difference so hence my conversion is without it.

rick decker wrote:

Yes you need to go over what you did and how you did it. I tried recovering some overexposed skies and did not have such luck. I don't think the DR of the shot was that great either as it was an afternoon shot with high overcast and the sun was behind me on the house which was gray and white.

One reason for the luck may be that he exposed at ISO 400.

If you look in the meta data you see that the CaptureISO then is 400, but the SensorISO is 200. This means that the data in the file is amplified as if it was taken at ISO 200.

So, the data in the file is one stop under exposed and therefore the highlight recovery is one stop extra.

Roland Karlsson wrote:

rick decker wrote:

Yes you need to go over what you did and how you did it. I tried recovering some overexposed skies and did not have such luck. I don't think the DR of the shot was that great either as it was an afternoon shot with high overcast and the sun was behind me on the house which was gray and white.

One reason for the luck may be that he exposed at ISO 400.

If you look in the meta data you see that the CaptureISO then is 400, but the SensorISO is 200.

How is this possible?

This means that the data in the file is amplified as if it was taken at ISO 200.

So, the data in the file is one stop under exposed and therefore the highlight recovery is one stop extra.

-- hide signature --

/Roland
X3F tools:
http://www.proxel.se/x3f.html
https://github.com/rolkar/x3f

Sorry if you've already covered this. Maybe an on-line instruction manual with some links to various places where things were discussed could be helpful. Does anyone know of such a resource? x3fTricks.com might be a good domain name for such a website. I guess the problem with such a site would be how quickly it might become obsolete, especially with new tools like Kalpanica/x3f Tools coming into existence. Certainly it would be a site that might be lucrative though, because hundreds (possibly even thousands) of people might become regular visitors. Companies like B&H Photo and Adorama might pay to advertise there, as would companies that make accessories for Sigma cameras, like those electronic viewfinders that were made for the DP Merrills.

Scottelly wrote:

Roland Karlsson wrote:

One reason for the luck may be that he exposed at ISO 400.

If you look in the meta data you see that the CaptureISO then is 400, but the SensorISO is 200.

How is this possible?

Now I have to guess what you mean. And I guess that you are wondering how it can be possible to store an ISO 400 image as ISO 200 data. Right?

1. We can take the Merrill first. Just like many other Sigma cameras, it does not have any AFE at all. The signal has the same amplification always. And this amplification corresponds to ISO 100. So, no matter what CaptureISO you set for Merrill, it will always be stored as SensorISO 100. Then, SPP and the OOC JPEG uses the CaptureISO and the SensorISO to scale the output with CaptureISO/SensorISO. So, if yoou set the camera at ISO 800, then SPP will scale all values in the RAW data with a factor 8. The main advantages with this method is simplicity and that you do not clip highlights when using higher ISO. On the contrary, increasing ISO, if you also expose less, makes it possible to recover more highlights.
2. The totally opposite idea is to have SensorISO equal to CaptureISO. So, if you set the camera at ISO 800, then the RAW data in the file is scaled a factor 8, probably by changing the analogue amplification of the signal before A/D conversion. This is how SD15 does it and most cameras on the market. Except that many cameras do not amplify the whole way analogue for high ISO. For high ISO, a part of the amplification is normally digital. So, ISO 3200 might be amplified 8 analogue and 4 digital. But, the result here is that you do not need to store ant CaptureISO or SensorISO in the meta data as the RAW data is already scaled. The advantage here is that you might suck some more info out of the darkness noise.
3. Quattro is a kind of compromise of 1 and 2. You do here amplify the signal (probably only analogue) but not the whole way. So, if you set ISO 400 (CaptureISO) then you amplify with a factor two, i.e. SensorISO 200. This is probably kind of optimal. You avoid clipping too much, still you can lure out some more out of the darkness. It is kind of clever, and Sigma needs to be clever because of their noise handicap.