New Sigma FP (L mount FF Bayer mirrorless) - About ISO6

[Guillermo Luijk]

Sigma has made an unexpected announcement: a new L mount FF Bayer mirrorless (and EVFless) camera.

Sigma announces ultra-compact 'fp' L-mount camera, teases full-frame Foveon sensor

I paid special attention to this:

"The camera has an ISO range of 100-25600, expandable to 6-102400. It has a fully electronic shutter and supports Eye AF and HDR. It can capture 14-bit DNG files and is fully compatible with Sigma's Photo Pro software. Its burst rate is an impressive 18 fps, though that's for only 12 shots."

I have a feeling that this ISO6 (and probably at a lower degree, the other ISOs in the ISO12-ISO50 range) can be quite interesting, much more than the fake "LOW ISO" found in most cameras. Let's put things together:

• ISO6 is 4 stops away from ISO100, too much to produce it by faking an ISO100 into ISO6 using metadata; highlights and even midlights would simply get blown away. So this ISO6 must be something REAL
• The shutter is fully electronic which means there can be no physical vibration, being this the natural shutter operation of the camera
• Finally, and contrarily to Sigma's usual slowness, it is aimed at being able to capture at a rate of 18fps, which means important readout and buffering capabilities

Putting all that together, I think this ISO6 is an averaging of 16 consecutive ISO100 shots with 1/16th the exposure time. Continuous vibration-free shooting means nearly having a real ISO6 (maybe moving light sources get a bit discontinued at times). What's this ISO6 useful for?

• Of course for long exposure photography (ISO6 means 16 times longer exposure time than ISO100)
• But also a Dynamic range enhancement of 2 entire stops for tripod applications (every x4 in captures means doubling SNR and hence gaining 1 stop of DR), which if you can afford the exposure time will basically mean nearly HDR RAW files

The procedure can be done with any camera, but probably with shutter vibrations and more importantly: not in a single output RAW file (I'm assuming the ISO6 setting will work on RAW here, but I see no reason why it shouldn't).

Regards

[DP]

but probably with shutter vibrations

nowadays few (if any ?) cameras do not have rolling electronic shutter as for not having a single raw - well - off camera software + off camera processing power might be a better option if you chase some better results... also EFCS does well against shutter shock (if you are using non continuous shooting - most, if not all cameras will switch EFCS off when shooting continuously) and using it might give you (for many cameras) raws with less noise (in deep shadows) vs using rolling electronic shutter in the first place... but of course in camera is convenient

[Dan Wells]

If it is (as widely speculated) the ubiquitous Sony 24 MP sensor inside the new FP, there's an easy way to look at the probable rolling shutter effect, even before having cameras in our hands... Set an A7III, a Z6 or an S1 to full electronic shutter. That's probably going to be what the FP is like all the time. I don't have any of the three around to try it...

[sbay]

The procedure can be done with any camera, but probably with shutter vibrations and more importantly: not in a single output RAW file (I'm assuming the ISO6 setting will work on RAW here, but I see no reason why it shouldn't).

I've tried this with my Sony with the app that gives a single raw. It's a great idea but often can result in visible artifacts due to the gap between shots. As a result I don't actually use it very often. I'm guessing the FP implementation will have much shorter delay between frames so might be more generally useful.

[Guillermo Luijk]

I've tried this with my Sony with the app that gives a single raw. It's a great idea but often can result in visible artifacts due to the gap between shots.

Do you see artifacts in fast moving lights (cars, planes,...) or even in regular subjects? would be nice if you show some example. I've done tests with moving water and relatively long shooting gaps to mimic ND filters and the results were great.

Regards

[sbay]

So if I want to flatten water completely, it usually works just fine for me. The problem I have is with slow moving objects where I want an exposure time of maybe 30s to a minute or two. I initially thought that with enough averaged frames (say 8-32 or so) any gaps wouldn't be that noticeable. Sometimes it works just fine but other times I get some artifacts:

Ex 1. Grass blowing in the wind. Base exposure 0.6s with maybe 16-48 shots averaged (I'm not sure because smooth reflection doesn't record the number of frames averaged in the metadata). I've had this work before with grass/leaf motion at this base exposure

Ex 2. Leaves blowing in the wind. Base exposure 0.1s. This was a shorter base exposure than I wanted to use but (didn't have my ND filters with me)

Ex 3. Slowly walking people. Base exposure 1s. The lights from the amusement park rides involving faster motion look ok.

Ex 4. Ocean waves. Base exposure 1s. I didn't use enough replicates here and there's some kind of weird interference pattern on the surface.

Probably I could have avoided some of these issues by using a longer base exposure or more replicates. But I didn't always have ND filters on me and/or wasn't sure how many replicates to use.

I'm using the smooth reflection app which has a noticeable delay between shots (requires mechanical shutter).

[capital]

If your assessment is correct that they will stack frames, I hope they are able to keep it as a raw mosaic. I had thought I might be able to do the same with Canon's 5DM4 with the multiple exposure facility, however it appears to create some type of re-wrapped 'fake raws' whose demoisaiced file's white and black points have been clipped--rendering the raw files worse than stacking after the fact.