Sigma Fp review and interview / Cinema DNG RAW

[Andrew Reid]

In my opinion the camera industry could learn a lot from the open source community.

Having so many exclusive patents and licenses really hurts the entire industry.

Every single company would benefit from an open source compressed RAW codec.

I think it is also what keeps us from better integration of the external solutions as well. We are never going to progress from a HDMI cables, because as soon as an external recorder connects more directly with a camera body, it probably counts as internal recording and thus violates the RED patent.

It stops more modular designs.

I would love to know what ideas and products RED's patent has killed off and whether this actually makes RED money in the long term.

I don't think it does.

Fair play to Mr Nattress. His work deserves long-term, rich reward.

But come on. It's time to push forward together as one industry.

[Lars Steenhoff]

If you need power and also ssd at the same time, this is a nice solution

 

[zerocool22]

On 6/26/2020 at 1:23 AM, rawshooter said:

It's not better with ProRes RAW.

Problem with BRAW is that it requires an encoding FPGA - so we'll likely never see recording internally except by BM cameras. 

Hows that? Why cant other companies implement an fpga? (Kinefinity and zcam should def look into this)

[Andrew Reid]

Partly because they suck batteries dry in minutes.

[stephen]

Sigma FP caught my attention. It is more compact and has a better form factor than BMPCC 4K. You can add a small lens and put the camera on something like Feyu Tech G6 Max and have a very lightweight gimbal setup. And can take much better stills. It is tempting as a travel cinema camera. As bjohn stated in another thread will wait and see what Sony and Canon will offer, but Sigma FP suddenly climbed on the top of my cinema camera wish list. 🙂

Found a very interesting article by Timur Civan. He proposes a workflow in Davinci Resolve which can extract 1-2 stops more dynamic range from Ciname DNG and claims  Sigma FP image quality is on par with Panasonic S1H. And those extra stops are in the highlights.
https://timurcivan.com/2020/06/an-examination-of-sigma-fp-raw-workflow-and-how-to-get-the-most-from-the-fp/?fbclid=IwAR3ga9HcmrnVdSr1R9V4Vb18_A5_BUsjeLYIPV33MF-2lic1sdgIDoKrb1U

After all both cameras share the same Sony sensor.

Basis of Timur’s method is the trick Juan Melara shows in one of his videos: Insider Knowledge - A better way to grade Ursa Mini CinemaDNGs

We had a discussion about it in another thread with Kye. It looks my understanding of this video is quite correct. Basic idea is to choose a larger color space and gamma when debayering RAW footage and specifically Cinema DNG in Resolve  (camera settings part). For the rest Kye was absolutely right. Once in Resolve any color space transformations have no impact on dynamic range.

Having in mind that with SlimRaw you can compress Cinema DNG 3:1 (lossless compression) and even more 5:1, 8:1 with lossy compression (similar to BRAW), all you need is just 1,2 more SSDs in the bag. Blackmagic Video Assist recorders would be useful in a more professional environment. For travel SSDs would be more than enough.

Will play with some Sigma FP footage from the net. Would be grateful if somebody knows where can download more RAW footage from Sigma FP.

[bjohn]

5 hours ago, stephen said:

Found a very interesting article by Timur Civan. He proposes a workflow in Davinci Resolve which can extract 1-2 stops more dynamic range from Ciname DNG

I watched his Vimeo walkthrough of how he grades Sigma fp footage, which was interesting and weird at the same time. His method for adjusting white balance is like nothing I've ever seen before and I can't wrap my head around it. I can't see how it could possibly work but I guess he manages. Normally I color-balance using the curves while monitoring the waveform with all channels displayed, or else I monitor the RGB Parade and use the color sliders (rather than the wheels). He does it entirely using the puck in the Gain wheel.

Anyway, I too am intrigued by the fp and have seen some inspiring footage from a few people (and a ton of uninspiring footage from others; I think there are maybe 3 or 4 clips on Vimeo that I like). Same goes for stills; most of the stills I've found on flickr or fredmiranda.com are unimpressive but the exceptions are really great and give you a sense of what this camera is capable of in the right hands.

It ticks a lot of boxes to me. I never use autofocus and have little need for IBIS, so those are not concerns for me; my only concern in the stills department is the electronic-only shutter and the potential for rolling-shutter artifacts on fast-moving objects. Among other things I take photos of traditional musicians and I worry what would happen with a moving fiddle bow, for example, especially in low light.

[BTM_Pix]

24 minutes ago, bjohn said:

My only concern in the stills department is the electronic-only shutter and the potential for rolling-shutter artifacts on fast-moving objects. Among other things I take photos of traditional musicians and I worry what would happen with a moving fiddle bow, for example, especially in low light.

Its fair to say the electronic shutter doesn't always fair brilliantly.

This train was only pulling away from the platform rather than speeding through it and the distortion is obviously, erm, "somewhat" apparent.

 

The bigger concern you might have in your application though is with LED stage lighting as the banding can be pretty horrendous as this quick snap of a sign in a lift shows. 

Suffice to say, there were no yellow and green stripes on the printed page !

 

[bjohn]

8 minutes ago, BTM_Pix said:

The bigger concern you might have in your application though is with LED stage lighting

Yikes, this does look like a deal-breaker. There's an informative article about this issue in Sony A7-series cameras here: https://phillipreeve.net/blog/limitations-of-the-electronic-shutter-function/

The Sonys at least have a mechanical shutter with electronic as an option. The Sigma only gives you electronic and this confirms my fears that this won't work for me. Bummer.

[BTM_Pix]

To be fair, the lift example really is as extreme as its going to get so don't take that as an indicator of its general performance every time it sees an LED light.

I have a Philips Hue light strip which is usually kryptonite for electronic shutters so give me 10 minutes and I'll come back with something.

 

 

[BTM_Pix]

50 minutes ago, BTM_Pix said:

So give me 10 minutes and I'll come back with something.

By which I mean at least 10 hours as something has just cropped up and I'll do it tomorrow.

[bjohn]

12 minutes ago, BTM_Pix said:

By which I mean at least 10 hours as something has just cropped up and I'll do it tomorrow.

Haha, no worries and thanks for doing the test! The examples you provided above were enough to convince me that this is probably not the right camera for me; if I were only shooting outdoors it would be fine, but I do shoot a lot of stills indoors and more and more the lighting comes from LEDs, plus my cinema lighting is LED and if I wanted to take stills during a video shoot I'd be stressing about the possibility of banding. And I certainly don't need any more stress during video shoots. 😉

[BTM_Pix]

13 minutes ago, bjohn said:

Haha, no worries and thanks for doing the test! The examples you provided above were enough to convince me that this is probably not the right camera for me; if I were only shooting outdoors it would be fine, but I do shoot a lot of stills indoors and more and more the lighting comes from LEDs, plus my cinema lighting is LED and if I wanted to take stills during a video shoot I'd be stressing about the possibility of banding. And I certainly don't need any more stress during video shoots. 😉

As I say, the shot in the lift is a very, very extreme example.

The effects will vary greatly depending on the frequency of the light and the shutter speed so its not a constant.

Looking at the exif of the lift shot, it is 1/1000th of a second, which will exacerbate it greatly and the type of light is an unknown variable.

I'll do a couple with just a regular vanilla LED video light as well so you'll know not every day is a yellow and green band day with this camera!

More tomorrow....

[imagesfromobjects]

2 hours ago, bjohn said:

Yikes, this does look like a deal-breaker. There's an informative article about this issue in Sony A7-series cameras here: https://phillipreeve.net/blog/limitations-of-the-electronic-shutter-function/

The Sonys at least have a mechanical shutter with electronic as an option. The Sigma only gives you electronic and this confirms my fears that this won't work for me. Bummer.

Fair points, but I think the only time that you'll see rolling shutter effects is with vehicles and things like tennis balls in flight and so on. I've used the fp for a few toddler events and kittens and whatnot and haven't seen any rolling shutter, so I doubt you'd see it in shots of a fiddler.

 

The banding issue can be mitigated by using shutter speeds that are multiples of 60 (1/30, 1/60, 1/125 etc) in the States or multiples of 50 in Europe. Gets tricky when you are working with mixed lighting or stage lighting, so it would be wise to test out different settings on site to see how it fares.

 

[bjohn]

17 minutes ago, imagesfromobjects said:

The banding issue can be mitigated by using shutter speeds that are multiples of 60 (1/30, 1/60, 1/125 etc) in the States or multiples of 50 in Europe

Yep, similar to what I do to avoid flickering in video in Europe vs. North America, but when I'm shooting stills I'm usually shooting manually in aperture priority mode (meaning the shutter speed gets set automatically) and I just glance at the shutter speed to make sure it's not too slow. I could switch to shutter priority in those cases to be sure I'm using a shutter speed that would avoid banding, but it would be simpler to use a camera that doesn't require me to consider workarounds like this. Unless the new Sony that'll be announced toward the end of this month fits the bill, I'll likely stick with my current solution of using separate cameras for stills and video. It's inconvenient but it works.

[Thomas Hill]

21 hours ago, bjohn said:

Among other things I take photos of traditional musicians and I worry what would happen with a moving fiddle bow, for example, especially in low light.

Steve Huff has some low light musician shots in his photo-focused review.

https://www.stevehuffphoto.com/2019/11/11/sigma-fp-camera-review-tried-one-then-bought-it-heres-why/

[Super8]

On 7/8/2020 at 8:44 AM, Thomas Hill said:

Steve Huff has some low light musician shots in his photo-focused review.

https://www.stevehuffphoto.com/2019/11/11/sigma-fp-camera-review-tried-one-then-bought-it-heres-why/

Steve raves about the FP. 

I've been doing some home work and have to find out the cost and advantages of the FP.

[TiiPii]

The statement from Sigma wrt. firmware 2.0 that Log is not possible in Sigma FP platform is really confusing because we know that it *can* apply curves to the raw signal from the sensor. That’s what all those different color modes are, just a curve. A perfectly fine way of implementing Log (or just a really flat) mode is to just apply a logarithmic curve. We know from Pawel Achtel, ACS, test (https://www.sigma-forums.com/forum/photography-discussions/photo-talk/174-deepdive-with-pawel-achtel-acs?p=175#post175) that Sigma FP has really wide color gamut, matching the entire gamut of Rec.2020. So applying a log curve over that gamut would give a really great range suited for preserving both the dynamic range and the tonalities.

So is Sigma really saying that they can’t add a new mode, or are they saying that they can’t even replace or modify an existing mode?

This is relevant of course only when you shoot MOV. If you shoot 12-bit raw then lack of log is not an issue. We have the whole raw 12 bits per pixel data available and that wide gamut can be fitted to whatever color space we want. A simple color space transform (LUT) will do. But it can be an issue with MOV. So as long as they could apply that curve for the signal before it gets crushed for Rec.709 or whatever, it would work great.

Sigma, please, more information needed what’s the actual reason for the “we can’t do Log”. Users aren’t asking for more dynamic range from the sensor. Just to preserve the dynamic range (and tonalities) it already has. Maybe they read this, maybe not… Maybe Andrew can ask Sigma directly?

[rawshooter]

9 hours ago, TiiPii said:

The statement from Sigma wrt. firmware 2.0 that Log is not possible in Sigma FP platform is really confusing because we know that it *can* apply curves to the raw signal from the sensor. That’s what all those different color modes are, just a curve. 

It's not just that. A proper log profile also has a wider color gamut. If your log curve remains within Rec709, it won't make much sense (and will cause color banding when translated into non-logarithmic Rec709).

It looks as if Sigma has been unable to implement a wider color gamut with its current camera hardware/ASIC. So it was sensible of them to pull the promise (rather than delivering just a Rec709 profile with a log gamma).

[rawshooter]

On 7/10/2020 at 10:41 PM, Super8 said:

Steve raves about the FP. 

He tested it one day after it came out, and only as a still camera for adapted manual rangefinder lenses with its more Instagram-y color profiles.

Which resulted in the Sigma fp having basically two user bases: CinemaDNG video shooters, and the vintage lens + stylized camera profile stills photo crowd.

But I think of him as just another industry shill and paid influencer.

[TiiPii]

1 hour ago, rawshooter said:

It's not just that. A proper log profile also has a wider color gamut. If your log curve remains within Rec709, it won't make much sense (and will cause color banding when translated into non-logarithmic Rec709).

It looks as if Sigma has been unable to implement a wider color gamut with its current camera hardware/ASIC. So it was sensible of them to pull the promise (rather than delivering just a Rec709 profile with a log gamma).

Agreed, having Rec.2020 baked into the MOV would solve the problem. Some users might have still found the camera a bit more appealing with a log/flat profile in 10-bit MOVs even if it's in Rec.709. (spec-wise at least). But Sigma is redeemed of course with the 12-bit raw.