Sigma has created one of the best cinema cameras on the market with better dynamic range than we thought, but at what cost?

Dynamic range

In my initial review of this camera, I was quite harsh in tone particularly with regards to the dynamic range of this camera. Through multiple Xyla tests, it has been shown that the dynamic range changes considerable as you change ISO. At 100 iso, you have unbelievable low light performance. CineD has posted tests that show that even at -6 exposure you can have a clean image when bringing it back up in post. You probably don’t want to film at this ISO most of the time because you have a dynamic range of 11. Most people would recommend shooting at 800 ISO as you get around 12-13 stops of dynamic range in post when you convert to Blackmagic colorspace. But lowlight performance suffers a bit, which is still better than any camera I’ve seen including Sony when underexposed. Notice the 19.1 patch range, it’d be noisy and unusable, but it’s still very impressive how the sensor is able to capture more dynamic range than anyone would expect. From personal experience, it feels cleaner and better than most camera footage I’ve worked with despite the numbers suggests, so let’s dig deeper into this mystery.

The DR is comparable if not better than some of the Red cameras like the Dragon X. I’ve noticed when pixel peeping the the noise profiles are different with Red having a chunkier luma noise whereas Sigma’s noise profile has less luma noise but much more chroma noise (color). Not to be a Sigma apologist, but chroma noise tends to be easier to get rid of in practice and is usually less noticeable for normally exposed scenes.

The FP has exceptional low light performance, but don’t expect to be raising shadows gracefully despite indeed being a vampire camera. It has tremendously low noise overall, shifting most of the stops below middle grey. This results in the dynamic range shifting downward, which results in rich shadows, contrast and texture, but an arguably less attractive highlight roll off compared to Arri-like cameras. If you like this look, you will describe it as heavy and substantial. If you don’t, you’ll describe it as horror and grungy. The same argument can be made for Red and Arri cameras. After minimal noise reduction and utilizing BRAW colorspace, you can get to 13 stops of usable dynamic range on the FP (SNR 2).

Sigma FP dynamic range after minor noise reduction
Gerald Undone - https://www.youtube.com/watch?v=Zsz3wto6lNk&t=534s&ab_channel=GeraldUndone

Xyla tests don’t tell the whole story. If I were to ask you if noise in the shadows or highlights bothered you, which would you say? I think everyone would agree that noisy shadows are worse, especially since it looks pixelly after video compression when uploading online. Noise in the highlights like the skies or bright lights is much less offensive and can even feel like pleasing grain. The FP has these cleaner shadows compared to other cameras, which I think with regards to noise performance makes it very attractive as a low light camera. It basically sees in the dark, probably the best low light camera on the market I’d argue. Sony is slightly better only at its second base ISO of 12800, but you have so much in-camera NR and color problems not to mention worse image texture.

So the FP does actually have very good dynamic range and you can realistically expect around 13 stops in practice for final output, which is actually similar if not better than other RAW cameras like Nikon, Blackmagic and some REDs. It also has an extremely accurate color science better than the Sony Venice, which makes mixed color situations easier to separate in the color grade, extremely relevant for low light scenarios.

Workflow

Despite being such a powerful camera, the workflow is cumbersome for most productions and makes it unideal for professional work. There is only micro HDMI after the usb slot is used up for data, so you won’t have secure connections, though I have been able to use SDI through a SmallHD monitor (I have also connected to wireless transmission systems feeds with the micro HDMI on a gimbal). Because exposure monitoring is inaccurate and also has low resolution, producers, directors and gaffers on set might not be able to have a clear picture of what the final image looks like. You’ll have to constantly reassure them it will look good since only you will understand the final image output potential of this camera. This is because the output image quality to any monitor has less bits than the actual image recorded, a hardware issue. Essentially, the image looks much softer and less detailed, more like 720p, which makes critical focus more difficult. The solution would be recording externally to a Atomos Ninja in BRAW, which would have the same resolution as the LCD. I think you need at least a 7in monitor for the FP due to this obstacle because this resolution hit makes a 5in not easy to rack critical focus. The focus peaking in camera isn’t reliable and can sometimes be misleading.

The highlights clip much faster on the monitor than what is actually recorded. The inaccuracy means that when monitoring, sometimes a face will have clipped white streaks, or highlights will be blown out, which is fine for the camera operator looking at false color, but confusing for everyone else on a set. Sometimes you and the gaffer just need to visually see what the lighting looks like through the monitor and exposure tools like false color are not always enough to gauge the artistic qualities of the light.

There is a director’s viewfinder and EL zone with this camera as well, fantastic tools, but the anamorphic modes are exclusive to this mode so you cannot de-squeeze natively when recording 12 bit 4k. Only h264 8bit anamorphic de-squeeze recording is possible within the directors viewfinder. CDNG raw is also a dilemma, as the file sizes are enormous and as large as Arri RAW. 50 minutes is basically 1TB of footage, which means you might need at least 2 SSDs because when you run out of storage it will take time to dump the footage on-site so you need a backup to keep rolling. Otherwise, depending on transfer speeds you can expect it to take an hour+ to dump 2TB (1hr50min) of footage so make sure production also has enough storage and time so you don’t have to idly transfer a single SSD over your lunch break (lunch break? HAH!). The RAW cDNG workflow is also not standard and most colorists and editors won’t be familiar with it. You’ll have to explain how to grade it, the reverse S curve of the FP’s contrast, the required lowering of highlights, the strangely inaccurate WB measuring in the raw tab, I’ve created free Rec709 LUTs for the FP for this exact reason. Essentially the default image of the FP is overly contrasty so you have to raise the mid-tones to de-contrast and then bring back down the highlight roll off.

Clever Ghost’s Sigma FP Grading Guide

So we’ve established that the images from the FP look comparable to the dimensionality of Red and Arri, how do we get there? This is my process of working with 12bit 4K cDNG. It is really a detailed codec with tremendous detail and depth. 1080p still looks nice almost filmic with it’s lower resolution, but there is sometimes noticeable line-skipping which doesn’t look great, also 1080p upscaled to 4K still has problems with banding for YouTube.

So here is step by step how I approach most Sigma footage.

Please note: my guide is developed for my taste, lenses and exposure preferences, which may not always suitable for everyone. It’s may not be the most proper or translatable method, but it is certainly the easiest and is worth always as much as you like the results. This is for simple shooters that don’t need to be professional colorists and spend all their time learning 20 node trees.

7. Increase the saturation (I recommend subtractive saturation with HSV saturation or color slice) and then adjust WB and tint to your liking.

A super simple process to get a very decent Rec709 base image utilizing 3 nodes. There is much more you can do with this image but I often believe that simplicity is divinity. For scenes with less than ideal lighting, you may want to do more with the grade to make it visually interesting. I’ve also made this power-grade available for free in the Discord server.

Rigging

Rigging has many options as this has become somewhat of a cult camera with many third parties wanting to make the best out of it. DPL (Dark Power Laboratory) is my choice as they have created an attractive half cage with an integrated SSD so you don’t have to worry about dangling cables. But depending on which M.2 NVMe SSD you buy for your enclosure, it may make considerable noise while recording and it sometimes has a beeping sound that makes it into the scratch audio. It also just barely blocks the battery door, you can still switch batteries with the SSD in, but it will be a strain on the battery door. If you choose the smaller integrated SSD cage, I would suggest the medium size as it still allows an option to access to the micro HDMI, which can barely fit a tiny right angle micro HDMI cable. Although micro HDMI is not a secure port, I have had essentially 0 problems with performance aside from the occasionally disconnect for 1 second. You probably do need to buy the optional EVF or a monitor when recording video as it’s really difficult to focus peak based on the LCD. I would not trust the focus peaking on this camera, it’s really hit or miss.

Final thoughts

Despite this sounding like the worst camera of all time, it is one of the most immersive images makers along with Red and Arri. Data is a practical issue as you’ll need Slimraw compression software, DAS, NAS or all 3. You could also save on data by shooting 1080p for a pleasant nostalgic film look, which can even be recorded to a memory card. 8bit 4k recording also saves space but you can only record in 5 seconds bursts to a card, however the dynamic range hit makes it so I won’t recommend it. Pictures are also detailed, though boring, I think warm gold, blue and teal, neutral, black and white are the only usable color emulations, the rest do weird things to skin tones. I always use neutral, which can pass as professional photos on par with Sony. Also the menus are straightforward but have some hidden Narnia settings, which you can only access by hovering over them. It’s a shame that they are not really making L to EF adapters with rear NDs, it’s mostly L to PL like the Mofage. Autofocus is decent, can only use it for a talking head maybe, nothing with movement.

This camera is truly a glass cannon of image quality with likely the worst workflow of any modern camera on the market. But if you can look past all these issues, which I can for my work, you are left with an image on par/superior to Red for basically $1000-2000 USD. If you want to make personal movies, there’s likely no better camera, but for professional commercial work, you’ll have to make lots of compromises and grade footage yourself. It may not have the attractive highlight roll off of Arri and Arri copycat cameras, but it has a distinctive look with strong micro-contrast and texture. This is the one thing you cannot emulate in post and is a majorly important quality when looking for a cinema camera. It downsamples from 6k to 4k and still has mediocre rolling shutter around 20ms (10ms for 1080p) but the downsample looks nice and creates a softer image, which I think pairs well with all the sharp modern glass today. This is all to say, the Sigma FP is the closest thing you can get to a modern Hollywood movie in terms of image quality. Other cameras like Blackmagic will get very close in look and color, but not in texture and depth. I genuinely think the Sigma FP is the best camera if you are looking for the ultimate image quality on a budget.