Sigma FP

[Ryan Earl]

14 hours ago, TomTheDP said:

It would be nice of course if they had included a nice log curve.

I've been experimenting with the Sigma FP as a quasi mini replacement for an 1st generation URSA Mini 4.6K shooting DNG.  My URSA has been declining in health with buttons not responding and the screen failing after 6 years. 

With the URSA Mini 4.6K I have an ingrained workflow with it's Gen 3 color science and the BMD Color Space and BMD Gamma.  The primary colors are accurate or at least what I've come to rely on.

Then with Sigma FP I transform the DNGs via the CST tool in Resolve from 'P3 D60' and 'Linear' to 'RedWideGamutRGB' and 'RED Log3G10' then transforming to REC709.  This gives me the feel of the Lift - Gamma - Gain tools that I like to use to render the color, contrast and saturation. 

Just in a few available light table top tests I was surprised at how well both files matched in color rendition.  I had seen some tests where the reds had shifted too orange on the Sigma FP (especially recording in BRAW and ProRes RAW) with identical ISO, white balance settings.  They do shift in my example but neither is perfect out of the box.

I attached two log like files below.

 

Sigma FP ISO 800 WB 5194 TINT 13.79

URSA Mini 4.6K ISO 800 WB 5194 TINT 13.79

[Ryan Earl]

16 minutes ago, Ryan Earl said:

Just in a few available light table top tests I was surprised at how well both files matched in color rendition.  I had seen some tests where the reds had shifted too orange on the Sigma FP (especially recording in BRAW and ProRes RAW) with identical ISO, white balance settings.  They do shift in my example but neither is perfect out of the box.

I should clarify that I was wondering if they were shifting to become more accurate or less accurate on the Sigma FP.  I think the Sigma FP's color relationships are more accurate (even more so than the URSA 4.6K) viewing the files without further adjustments inside the Linear to RED to REC709 transform I described above vs other decodes and transforms.

[TomTheDP]

3 hours ago, Ryan Earl said:

I should clarify that I was wondering if they were shifting to become more accurate or less accurate on the Sigma FP.  I think the Sigma FP's color relationships are more accurate (even more so than the URSA 4.6K) viewing the files without further adjustments inside the Linear to RED to REC709 transform I described above vs other decodes and transforms.

Interesting, I plan to shoot using CDNG. This is one of the reasons I am going to buy the FP, seems to have excellent color rendition.

[Ryan Earl]

Another thing I've noticed is that the colors shifting in DNG has to do with the Color Mode "OFF"  setting.  The images I attached earlier were done with the Color Mode set to "Standard" as viewing lut.  It might have been stated earlier in the thread, these modes do not affect the CDNG development in Resolve, so not matter what setting you get the same CDNG.  Though when setting the Color Mode to "OFF" you get a slightly shifted primary that corresponds to the same raw output in the 4K DCI BRAW. 

So if you're only ever shooting DNG it seems best to try both and see what you prefer.

[TomTheDP]

2 hours ago, Ryan Earl said:

Another thing I've noticed is that the colors shifting in DNG has to do with the Color Mode "OFF"  setting.  The images I attached earlier were done with the Color Mode set to "Standard" as viewing lut.  It might have been stated earlier in the thread, these modes do not affect the CDNG development in Resolve, so not matter what setting you get the same CDNG.  Though when setting the Color Mode to "OFF" you get a slightly shifted primary that corresponds to the same raw output in the 4K DCI BRAW. 

So if you're only ever shooting DNG it seems best to try both and see what you prefer.

Yeah that is my plan. I heard that about shooting BRAW, is it the same case with Prores Raw?

[Ryan Earl]

ProRes Raw also has the color shift.  It seems by design with a look imprinted into the footage or a kind of LUT.  So you have to want to work with the colors you are getting from it I think, not sure if there is a way around it.

 

 

[TomTheDP]

On 4/17/2022 at 8:49 AM, Ryan Earl said:

ProRes Raw also has the color shift.  It seems by design with a look imprinted into the footage or a kind of LUT.  So you have to want to work with the colors you are getting from it I think, not sure if there is a way around it.

 

 

micro HDMI is a no go for me anyways. I am willing to just shoot HD 8bit RAW if I need smaller files.

[webrunner5]

I am amazed that the 1DC has a Mini HDMI port, what the hell. Better than a Micro but this was a Pro Cine camera, a $15,000.00 one at the start!

 

[TomTheDP]

1 minute ago, webrunner5 said:

I am amazed that the 1DC has a Mini HDMI port, what the hell. Better than a Micro but...

Yeah i guess mini might be ok. It seems like every camera company is in on a big scheme to gimp every camera LOL.

[webrunner5]

I feel less confident using the 1DC with the mini than I do with the OG BMPCC with the micro. The cost of the 1DC would be crazy I bet to fix. I use my Zacuto on the flash mount when I am shooting in C Log. No AF in that mode.

[TomTheDP]

34 minutes ago, webrunner5 said:

I feel less confident using the 1DC with the mini than I do with the OG BMPCC with the micro. The cost of the 1DC would be crazy I bet to fix. I use my Zacuto on the flash mount when I am shooting in C Log. No AF in that mode.

I see a lot of used BMPCC with broken micro HDMI. It's actually why I haven't repurchased one of them. The screen on the OG Pocket is also total crap. I have heard the sigma FP screen is pretty nice, aside from its lack of tilt. If I get one I will buy the tilt modification kit.

[Ryan Earl]

The HDMI on my BMPCC broke off somewhere in 2014!   I think the FP is an interesting replacement option because of the 12 Bit CDNG HD to SD cards vs the 8bit UHD recording in camera.  I've been recording mostly with the USB C port to a Samsung T5 so now I'm worried about that!  I bought the SmallRig cage but can't find their cable clamp attachments that would fit.  

The Sigma made attachment does work pretty well to lock in the HDMI cable though and comes with the kit that includes the 45mm lens. I've been monitoring with the BMD video assist 12g.  

The zebras on the Sigma and the video assist have me thinking I'm clipping 2 stops into the highlights  when the Sigma's exposure indicator is at "0" which are actually mostly recoverable in Resolve.  I've been using ISO 800 exposing using a gray card with the false color on the BMD Video Assist. So I get green on the gray card, think I'm clipping, then bring them back, though I have to make sure it's not clipping more than 2 stops.  

I'll have to read through the threads on here to find an easier way to expose it, but for now I'm fine even with a few hard clips and using Resolve's soft clip tool to roll them off a little better.  

[Anaconda_]

6 minutes ago, Ryan Earl said:

I'll have to read through the threads on here to find an easier way to expose it, but for now I'm fine even with a few hard clips and using Resolve's soft clip tool to roll them off a little better.  

I've just set the zebras on the VA to a lower % Expose so the highlights hit the zebras and then knock it down a touch. Seems to give me good results.

 

[markr041]

I shot in 8bit 4K RAW by accident. How bad is it?

 

[markr041]

The Lumix L 24-105 f4 lens AF works well on the Sigma fp, and it is near parfocal (12bit RAW 4K videos):

 

[webrunner5]

Very well done. Thanks for the footage.

[kye]

I'm trying to find a 1080p 12-bit workflow up to 120p from the FP into Resolve, but using a compressed codec to limit file sizes.

The FP lists 12-bit 1080p up to 119.88p on the HDMI Output, so that seems good:

The BM Video Assist units don't list any frame rates beyond 60p, even in lower resolutions:

Also rather troubling is that they list the HDMI as 10-bit.  Does that mean that BRAW is 10-bit from these units?

The Atomos Ninja lists 12-bit so I'm assuming we can get 12-bit into the unit:

Then the codec challenge is there because Resolve doesn't support Prores RAW, and the Ninja doesn't support the 12-bit Prores flavours.  

It does support DNxHR HQX which I believe is 12-bit (link) and I believe Resolve supports the format (it will render proxies in the format so seems good?) but DNxHR is UHD only - the 1080p version is DNxHD (the last letter is different) and those seem to be limited to 10-bit.

There is a converter to convert from Prores RAW to CinemaDNG which Resolve will read, but I'd like to avoid the transcoding step if possible.

Any advice would be appreciated.....

[Anaconda_]

@kye

I just checked, with a VA I can record Braw 1080p 119.88fps. 

In Resolve, a clip I shot at UHD 25p is said to be 16bit... but I don't think that's right. This file is straight out of the Video Assist, recorded in Braw Q5. 

 

In an app called SCOOP! - which tells you metadata, it says the bitrate is Variable.

 

I'm not sure exactly what the numbers are, but in real life, I love Braw out of the Sigma. It's my go-to camera at the moment.

EDIT: 

Interestingly, a clip from my P4k also says 16bit in Resolve... 

[kye]

3 hours ago, Anaconda_ said:

@kye

I just checked, with a VA I can record Braw 1080p 119.88fps. 

In Resolve, a clip I shot at UHD 25p is said to be 16bit... but I don't think that's right. This file is straight out of the Video Assist, recorded in Braw Q5. 

 

In an app called SCOOP! - which tells you metadata, it says the bitrate is Variable.

 

I'm not sure exactly what the numbers are, but in real life, I love Braw out of the Sigma. It's my go-to camera at the moment.

EDIT: 

Interestingly, a clip from my P4k also says 16bit in Resolve... 

Thanks, that's really useful.  I would have been very surprised if I could do 4K60 but not 1080p120!

I have seen various references over the years of bit-depths being reported incorrectly - 12-bit being recorded as 14 or 16 bit, and 14 bit as 16 bit so it might be a common thing.  As soon as you go to 10-bit then you need (in theory) two bytes to contain the value, which is 16 bits, so maybe it's saying that the container is 16-bits, but of course the data will be limited to whatever bit depth the sensor/camera is capturing.

If I go this way I would prefer a BM solution as the integration with Resolve will be perfect and the bitrates will be much more manageable.  One article said that 3:1 BRAW was the same as Prores HQ, so I'll be able to get equivalent bitrates but with greater bit depth, plus on the 120p shots I can use as little as 12:1 and get roughly the same bitrates as 24p.  This means 3:1 1080p would be ~180Mbps vs FP internal 610Mbps, and for 100p I can go from the FP 2,530Mbps to as low as ~180Mbps.  You can often get away with more compression on slow-motion footage as things don't change much from frame to frame but it's nice to have the option.  I don't think that Prores RAW has as many compression ratios available - just the two.

I note the VA can record to SSD via USB, so that might be handy too.

[GUDZON]

Hello everyone, I have owned Sigma fp for a year now, and I want to say that this camera will never cease to amaze and delight me. I am not professionally engaged in this video filming and in connection with this I would like to clarify something. In this video link below, I've interpreted the internal CDNG (12 bit), perhaps a little differently. And I know that from the point of view of the "science of color" this is not true. But I really liked the result. Trying to process the same files that Ryan Earl suggests (via CST in RED Log3G10) - I get a really accurate picture, but it does not convey a sense of color depth. Do I need to use additional presets after conversion?

Video here (Youtube)