Sigma's secret weapon - SD Quattro review, an incredible filmic 8K timelapse tool with infrared capabilities

[Grégory LEROY]

1 hour ago, Grégory LEROY said:

Yeah, the red are so rich and vivid, bayer looks faded in camparison..before buying it, I still have to figure out how to build a good light set up with sigma (radio trigger? any remote control? which monolight and speedlight?), and to know if it's possible to bring the magnification to the eye caus I haven't succeed at sigma shop. 

Yongnuo flash triggers don't work with the dp quattro because of the lens bulge, do you know if the GODOX – X1 work?

[maxotics]

3 hours ago, Grégory LEROY said:

Yeah, the red are so rich and vivid, bayer looks faded in camparison..before buying it, I still have to figure out how to build a good light set up with sigma (radio trigger? any remote control? which monolight and speedlight?), and to know if it's possible to bring the magnification to the eye caus I haven't succeed at sigma shop. 

I believe there's a zoom assist, probably buried.  As for radio control, you can get something like this https://www.adorama.com/us 934638.html?gclid=Cj0KCQiA7dHSBRDEARIsAJhAHwjCws--uxALtZ9Oln3j_78EvcqcNnWm4ZzbuMSq5-hBycBDGr18caEaAgYMEALw_wcB   

[Grégory LEROY]

2 hours ago, maxotics said:

I believe there's a zoom assist, probably buried.  As for radio control, you can get something like this https://www.adorama.com/us 934638.html?gclid=Cj0KCQiA7dHSBRDEARIsAJhAHwjCws--uxALtZ9Oln3j_78EvcqcNnWm4ZzbuMSq5-hBycBDGr18caEaAgYMEALw_wcB   

there's a zoom assist but we cannot move it to the border of the frame which doesn't make sens...this adorama link seems to not work in Europe.

I was watching the picture, colors are so deep, but in the same time, is it not too contrasty? Must be good for dramatic portrait. I'm really thinking about buying it but in the same time, why nobody use it in photo studio?

[webrunner5]

The link led to a Adorama add for a "Canon EOS Dedicated TTL Off-Camera Shoe Cord 2, 2' Long." Like 10 US dollars.

[maxotics]

7 hours ago, Grégory LEROY said:

I was watching the picture, colors are so deep, but in the same time, is it not too contrasty? Must be good for dramatic portrait. I'm really thinking about buying it but in the same time, why nobody use it in photo studio

I don't know if I mentioned this before, Bill Claff did a DR test of some Sigma cameras and found a DR of 6.  He was flamed off some forums for reporting those findings, which I believe true.  But they do NOT make a difference, at least to me.  We don't look at images beyond that range.  The flip side of the question is, aren't the Nikon images too low in contrast (too washed out?).  Nikon has the widest DR tested.  But again, the question isn't how much DR of the physical world a camera captures, but how well it can deliver sensor values into our 6 DR viewing space.  In my tests, when I've tried to match bayer images with Sigma images I have never been able to get the same saturation look.  The bayers always look slightly washed out in comparison.  That's to say, if you do get the Sigma camera and have the time to work each RAW image to match contrast, the Sigma will still maintain that 3D look--definitely against an APS-C sized sensor.

As for why no one uses it in a studio.  Let me be blunt.  A studio is for fast, efficient professional work where the end-user is almost never looking for the kind of quality that separates cameras.  Therefore, the Sigma is too slow.  Second, most photographers are technically weak; that is, they don't really understand what the camera is doing under the hood.  Third, they need to justify in their mind why their camera is already the best.  To be fair, as I've said, I would not use a Sigma cameras for professional work.  I'd only use it as a fine art photographer.  I'd DEFINITELY use it for portraits, but again, not for typical professional work--it's just too slow and time is money!   If you're just starting out and don't have the money for a D810/D850, the Sigma will give you a Medium Format look that I believe will impress people.  

I did some tests I did 5 years ago trying to match bayer to Sigma cameras.  I couldn't do it.  But again, if I don't know any better (like most people ), bayer images are fine!

http://maxotics.com/2012/11/26/sigma-dp2s-vs-sony-nex5-with-18-55mm/

  

[HockeyFan12]

5 hours ago, maxotics said:

I don't know if I mentioned this before, Bill Claff did a DR test of some Sigma cameras and found a DR of 6.  He was flamed off some forums for reporting those findings, which I believe true.  But they do NOT make a difference, at least to me.  We don't look at images beyond that range.  The flip side of the question is, aren't the Nikon images too low in contrast (too washed out?).  Nikon has the widest DR tested.  But again, the question isn't how much DR of the physical world a camera captures, but how well it can deliver sensor values into our 6 DR viewing space.  In my tests, when I've tried to match bayer images with Sigma images I have never been able to get the same saturation look.  The bayers always look slightly washed out in comparison.  That's to say, if you do get the Sigma camera and have the time to work each RAW image to match contrast, the Sigma will still maintain that 3D look--definitely against an APS-C sized sensor.  

This is a very interesting comment. 

I sort of laugh when DXO Mark assigns a good "landscape" score for 14+ stops of dynamic range (that I doubt are really there) as if we need 14 stops of dynamic range to shoot landscapes. Granted, the zone system asks for 10, but most color landscape photography (almost all of the best stuff) is shot on slide film, which has 5-6 stops of dynamic range. 

A printed photo will never have more than five stops of contrast (except with vantablack). So anything with a lot more dynamic range in the original scene will either look washed out or need to be compressed in processing or with an ND grad for a print. Even the zone system usually results in a rather tone-mapped print. (I never thought the zone system and tone mapping would work well in color–see Peter Lik's nicely composed but garishly processed photos. But then I realized it sort of can–see the Revenant, which is beautifully processed.)

The best looking (color) images I've seen, at least technically, are relatively "flat" scenes shot on 4x5 Velvia and viewed as slides on a light table–comprising four stops of scene dynamic range viewed backlit on an analogue display with as much contrast (display dynamic range) as an OLED and far better resolution and color. Looks hyper-real. Hyper-saturated. 

The Sigma doesn't get close. Not even close. But it does get a lot closer than most dSLRs.

(Fwiw, LED screens can display almost 10 stops of contrast, as backlit displays are much more contrasty than front-lit. OLEDs and plasma can display even more, but room ambience washes it out and you need a very high nit (10,000+) display to get the effect of a HDR display, which is even more stunning than the light table but not commercially viable.)

Furthermore, this gets controversial so take it with a grain of salt, but I had to grade footage from a high end 4k camera with a bayer sensor to match Kodak motion picture film. I used the vector scope to match chips on a color checker chart to set up a basic grade and found two things: even when everything matched (same saturation and hue) the film looked more saturated; and I could never match the vectorscopes perfectly because the graded video had a more diffuse point cloud around each color, the film had narrow peaks. I'm generalizing here, of course, but I believe Bayer sensors have inherent limitations. (So does Foveon, just different.)

[maxotics]

39 minutes ago, HockeyFan12 said:

Fwiw, LED screens can display almost 10 stops of contrast, as backlit displays are much more contrasty than front-lit. OLEDs and plasma can display even more, but room ambience washes it out and you need a very high nit (10,000+) display to get the effect of a HDR display, which is even more stunning than the light table but not commercially viable.)

Very interesting!  Thanks!  Now I understand why the published contrast ratios don't work in "real life".  

[HockeyFan12]

9 hours ago, maxotics said:

Very interesting!  Thanks!  Now I understand why the published contrast ratios don't work in "real life".  

I could be wrong. That was my just my impression when I saw a 10000 nit display as compared with other high contrast displays (my 2013 Panasonic Plasma; OLED displays at stores; the iPhone X). It was like looking out a window and I completely understood why HDR capture requires 15 stops or whatever. It's all there.

But for landscape prints studio prints I think six stops of DR is good enough. (Just imo.)

[Grégory LEROY]

On 10/01/2018 at 1:35 PM, maxotics said:

I don't know if I mentioned this before, Bill Claff did a DR test of some Sigma cameras and found a DR of 6.  He was flamed off some forums for reporting those findings, which I believe true.  But they do NOT make a difference, at least to me.  We don't look at images beyond that range.  The flip side of the question is, aren't the Nikon images too low in contrast (too washed out?).  Nikon has the widest DR tested.  But again, the question isn't how much DR of the physical world a camera captures, but how well it can deliver sensor values into our 6 DR viewing space.  In my tests, when I've tried to match bayer images with Sigma images I have never been able to get the same saturation look.  The bayers always look slightly washed out in comparison.  That's to say, if you do get the Sigma camera and have the time to work each RAW image to match contrast, the Sigma will still maintain that 3D look--definitely against an APS-C sized sensor.

As for why no one uses it in a studio.  Let me be blunt.  A studio is for fast, efficient professional work where the end-user is almost never looking for the kind of quality that separates cameras.  Therefore, the Sigma is too slow.  Second, most photographers are technically weak; that is, they don't really understand what the camera is doing under the hood.  Third, they need to justify in their mind why their camera is already the best.  To be fair, as I've said, I would not use a Sigma cameras for professional work.  I'd only use it as a fine art photographer.  I'd DEFINITELY use it for portraits, but again, not for typical professional work--it's just too slow and time is money!   If you're just starting out and don't have the money for a D810/D850, the Sigma will give you a Medium Format look that I believe will impress people.  

I did some tests I did 5 years ago trying to match bayer to Sigma cameras.  I couldn't do it.  But again, if I don't know any better (like most people ), bayer images are fine!

http://maxotics.com/2012/11/26/sigma-dp2s-vs-sony-nex5-with-18-55mm/

  

I have time, I only work for my own project (but I'm printed internationally), so I'm going to invest in Foveon.

Next week, I will try the SD Quattro with some old sigma lens like the 50mm f2.8 macro and the 17-70mm (I don't need hyper fast glass), I hope it's going to be easier to manual focus than on the dp.

[Waynes]

I have skipped through the article.  I was looking for a list of vertical color filter sensor makers.  In the past I've found one Russian firm, Canon, and Siemens might have had a Foveon technology licensed mobile sensor.  If anybody knows of a list, please let me know.

Now, there has been some issues with the Foveon.  Apart from the reds getting noisey and dim, as the light passes through the layers, some of ths color meant for deeper layers gets lost, resulting in that noise and unbalance. They compensate for this.  But the layers themselves are artifical cutoffs, not how the eyes overlapping layers work.  These are potential routes for inaccuracies.  You really need at least 5 layers to get it more accurate.  Now, the cutoff is likely soft enough to give you some overlap.  But two issues, accurately shaping each response layers to human vision is good but even more accurately recording values across the color ranges is better for software to get a hold of things more accurately.

Now, Foveon also has gone to a combined pixel in one color layer, something I don't want. The 4:2:0 of layering.  I don't see how it could be as accurate.

I have waited decades for video performance from these chips, a number have.  But it didn't come through and the market turned down.  Good video performance was going to be a seller of cameras.   It was just it was so poor back then, but even more reason to buy an X3 with good video performance back then.

There actually was a Foveon chip years ago that seemed to have aspect to pull 720p24 off sensor, which people could have jumped at.

But the issue with Foveon is that Sony got and bought vastly superior sensor technology outside of X3 itself (and are supposed to be developing X3 like technology).  Canon also has technology.  Their best bet is to partner with a company like the one that bought Aptina, that has a lot of cross licensed Sony technology, or samsung or Red.  Then they could do the sensors with updated X3/X5+ technology, with very good low light and HDR for Sigma cameras, high end security cameras, phones, and broadcast pro cameras, other video cameras and some pocket cameras.

Frankly I want a x7 (UV, 3 primary two complimentary, and IR).  Very useful for a range of things.  But for cinema x5-x7v to better match the human eye (x7v: blue, a spread accross the overlaps with green, and red).  Even a fullhd version would be useful.  But realistically 4k.  The data stream would be mixed down to 4:4:4 to keep the data rate down, even 4:2:2 or 4:2:0 consumer.  We can basically do away with the debayering and color grading steps in ENG and broadcast due to the increased accuracy over Bayer or film, with simple adjustments and auto color look settings.  There was a recent article, maybe on not a film school, about robot colourists doing a good job.  Outside of cinema, auto stuff to a desired look should be sufficient. The issue with Bayer, I see it produce some weird stuff in broadcast.  This means an undesirable grading step, or some auto approximation.  There was a time you would get shame for even mentioning using single chip in such environments (maybe due to complimentary single chip consumer cameras being so lousy).  But Bayer is better.  What is needed in broadcast, is no hassles accurate live results.  That is worth money.

 

[Waynes]

Is there a hack to overclock the sensor to output at video rates on HDMI at 720p, 1080p or uhd?  It would be worth buying a feild recorder to record this then.

[Waynes]

I take it back.  News of a Foveon video sensor patent:

http://thenewcamera.com/sigma-patent-new-foveon-sensor/

[Mattias Burling]

If you dont feel like hacking you can just buy a Sigma foveon that shoots video. But you won't get 720p

[Waynes]

I'm trying to find a Sigma contact form to ask what fps and live modes the HDMI has.  Also to suggest a firmware update to output on HDMI unaltered video downscaled to some mode, and binned resolutions for higher frame rates.  I imagine you should get at least 720p24 binned.

[jgharding]

Glad to see a review of a Sigma! They're still trying new things and haven't given up on Foveon which is amazing.

People often forget Bayer arrays ahve like half the resolution they claimto, and you need 8K to get true 4K and so on (I had to explain once to a graphics guy why generated graphics were twice as sharp as footage at a given resolution.

I used to have this DP3, and I used a wide angle adapter to get a wider view sometimes as it's fixed 50mm which is portrait on APSC

It has the most beautiful skin tones, I still think this is the most overlooked Portrait camera. Yes literally any movement above ISO 100 started to screw things up, but for portrait that was OK.

The downside is you couldn't process the raw in anything but their PAINFULLY slow and clunky software, which ruined the whole experience. I don't think this has changed AFAIK, So it went the way of eBay.

It's amazing how software can ruin a great camera.

[BTM_Pix]

5 minutes ago, jgharding said:

The downside is you couldn't process the raw in anything but their PAINFULLY slow and clunky software, which ruined the whole experience. I don't think this has changed AFAIK, So it went the way of eBay.

It's amazing how software can ruin a great camera.

Sigma now do a plugin for Photoshop

https://www.sigma-global.com/en/download/cameras/sigma-x3f-plug-in-for-photoshop/

The Quattro series can use DNG so its less of an issue than it was on the DP series.

Sigma PhotoPro has improved quite a lot from when I first started using it but, yeah, its still not exactly super fast.

Most of the time I just use the batch processor in PhotPro to save them to TIFFs and then edit them in Lightroom.

 

[jgharding]

5 minutes ago, BTM_Pix said:

Sigma now do a plugin for Photoshop

https://www.sigma-global.com/en/download/cameras/sigma-x3f-plug-in-for-photoshop/

The Quattro series can use DNG so its less of an issue than it was on the DP series.

Sigma PhotoPro has improved quite a lot from when I first started using it but, yeah, its still not exactly super fast.

Most of the time I just use the batch processor in PhotPro to save them to TIFFs and then edit them in Lightroom.

 

you know what, I enjoyed the DP3 so much the idea of using it photoshop makes me thinking of buying one again!