SIGMA fp Review – First Impression, Sample Footage and Lab Test Results

October 18th, 2019

Update: New firmware to combat some of the flaws we found is now available to download from SIGMA’s website by clicking here.

The SIGMA fp full frame camera is almost here and we were lucky enough to put our hands on a production model device. BUT, as we are so close to the official delivery date (October 25th), it means that the company is working hard on making further tweaking to the camera firmware, so the image quality you see above is NOT final as I was filming with a NON final firmware version. Eager to learn more? Hope you will find my SIGMA fp review useful. 

SIGMA fp – A true pocket size camera (Image credit: cinema5D)

We have been covering the launch of SIGMA’s first mirrorless camera that can shoot video since it was announced about three months ago. Personally I have a great interest in this product as this development has different aspects. To start with, it is an interesting move by SIGMA, a company which is well known for making quality lenses but when it comes to cameras, their Quattro line is aimed for photographers and those who want to explore the advantages of using the Foveon sensor for their photography leisure or assignments. Secondly, in what seems to be a crowded market, SIGMA saw an opportunity to shine by making something a bit different and I was keen to find out what it is. What can make this little (yet big) camera special? After playing with it for a few days, do I have the answer? Maybe not in full, but close enough to share with you guys my findings. 

SIGMA fp – The Philosophy

When Yamaki-san, SIGMA’s owner and CEO was first to introduce the camera, he mentioned “portability”, “scalability”, and “seamlessly” (the ability to change shooting modes between video and stills effortlessly). I can confirm that all those points were fully fulfilled. This small in size camera is very capable of producing nice moving images (depending on the recording mode you choose) and when it comes to scalability, I truly liked what I saw. It is very refreshing to see a manufacturer who takes the responsibility of offering a wide range of original accessories to its creation and not rely on third-party vendors. Speaking of which, the camera is being delivered with the “left side accessory” where the cold shoe attachment and HDMI protector resides. One thing that truly strikes me is the build quality of the camera and the accessories I’ve used. Both camera and Viewfinder felt like tanks and the attachment point between the two is seamless, meaning, it felt secure and robust. So much robust that one can hold and pick up the camera by holding the VF. On top, if you are in a need for a fast temporary solution for placing your external SSD, look no further. You can actually nicely gaffer tape it on top of the VF without fearing sudden drop.

SIGMA fp and Samsung T5 SSD gaffered on the viewfinder. (Image credit: cinema5D)

SIGMA fp – The Menu Structure

Camera menus, in general, are a sore subject in our industry… I can maybe count one manufacturer who does it right when it comes to simplicity and effectiveness. In this regard, SIGMA was sticking to their Quattro camera menu and as a first time user, I have to say that I had some difficulties to adjust. In a way, the menu structure is trying to be very straight forward but in reality, so many functions are buried deep inside under sub-menus. So plan to take some time to learn how to get the best out of it (here is a challenge. Try and find where to turn the “Zebra” on/off). When it comes to SHORTCUTS, the “QS” button might be your best friend. Try and customize it for your needs. As the camera does not have so many buttons in general, I would have liked seeing better usage of assigning shortcuts to them. One sample is the back wheel keys. Only two out of the four buttons can be assigned and this is a bit of a pity. While being on the subject of the back wheel,  I find it to be much too sensitive. In my case, it was assigned to control the shutter speed and unfortunately, I found myself turning it by a coincidence too many times to the wrong shutter speed value.

SIGMA fp QS menu – The easier way to get around. (Image credit: cinema5D)

Camera Picture Quality

The new SIGMA fp camera is capable of producing nice-looking images if shooting on CinemaDNG RAW format, no matter what bit depth you choose to record on. But let me start from the beginning. Currently the camera is capable of shooting in three different formats: MOV ALL-I (4:2:0, 8bit, Up to 420 Mbps), MOV GOP (4:2:0, 8bit up to 120 Mbps) and CinemaDNG (RAW recording format up to 2500 Mbps). In each recording format, one can choose different resolutions and frame rates. When choosing CinemaDNG, be prepared too work with a format that will “drink” the capacity of your recording media, but at least the camera is smart enough to allow the flexibility of using different bit depth externally (8, 10, 12 bit), and a single bit depth when shooting into an SD card internally (8bit). I can only hope that a more efficient RAW recording format will be supported in the near future and hopefully it will be one that doesn’t require an external recorder (the external SSD recording solution is working perfectly well and helping in keeping the small form factor of the camera too. I had great success using the Samsung T5 1TB SSD. My only griff was not buying the 2TB version)… Picture quality-wise, the outcome is mostly clean and pleasing to the eye. I write “mostly” as you never know when Moiré will surprisingly raise its ugly head. This is the result of the camera not heaving an OLPF in front of the sensor. (Optical Low Pass Filter. More information about the functionality of this filter can be found here). In addition, at times, I found the images coming out of the camera to be a bit too contrasty. This is why I chose to change my gamma settings in Resolve 16 to 2.6 while “developing” the RAW images in order to have a less contrasty image and more details in the shadows (without adding additional noise). Things get a bit more complicated when choosing to shoot in MOV format. Although ALL-I is good to have and 420 Mbps is a good benchmark (Albite 4:2:0, 8bit), the picture coming out of the camera is noticeably inferior to what we got used to in this category. Moiré can become very evident especially when moving the camera (again the absent of OLPF is very noticeable). For me personally it is a drawback as it hurts the overall flexibility of the camera. There was some discussion on the net about the logic of including 8bit CinemaDNG RAW recording option, and I guess one of the reasons for it is to allow higher video recording capability in a reduced data consumption (if comparing to 12bit), so if you find the MOV format to be not satisfying enough for your project, maybe filming in CinemaDNG 8bit will be a proper solution. Indeed I’m comparing apples to oranges, (MOV vs. CinemaDNG), but I’m raising this point as a possible, yet not perfect, solution.

A frame from the timeline – Be aware of possible Moire

SIGMA fp – Strengths and Weaknesses

Strengths: (In no particular order)

  • One of the main things to remember is that the SIGMA fp camera is priced at  $1899. A very aggressive and competitive price for a full frame sensor camera with RAW video recording capability. 
  • The SIGMA fp is an extremely compact camera. I love it for that reason as it allowed me to operate incognito everywhere.
  • Built quality is very impressive and in terms of scalability, it has a masterpiece design! As an example that indicates the camera built quality, there was a night when I was running with it and a sudden pouring rain surprised me. I’m happy to report that although the camera got COMPLETELY wet, I had no issue to turn it on and start shooting after wiping it with a dry towel. 
  • CinemaDNG RAW recording internally (SD card) and externally to a fast SSD is an extremely welcomed feature. I will always value such an option over using an external recorder whenever possible. 
  • The lowlight capability of this camera is exceptional. With a base ISO of 100, I find it simply amazing in regards to how high I could go and how usable the images from the camera in those high ISO values are. (On minute 4:38 I filmed Mai-san in the car with only the very small door light activated at ISO 25,600. Grainy but absolutely useable ).
  • Battery life: I didn’t measure how long one lasts (sorry, was busy filming), but it is not that I felt at any point that it is too short.
  • Camera getting hot while recording: I tried different shooting scenarios. Indoor, outdoor, interviews, and general image gathering. I didn’t really have the possibility to take it to the extreme, so some additional testing is required. (In those normal filming conditions I had no issues at all).
  • Nice to see that SIGMA added the ability to record audio at different levels on channels 1 and 1 if wanted.

SIGMA fp – What the eye can see (Image credit: cinema5D)

 

SIGMA fp – ISO 6400 (Image credit: cinema5D)

Weaknesses: (In no particular order)

Please allow me to separate between those functions that I think are missing and those that are included but could work better. Based on offerings by other respected manufacturers and user requests, this is my list concluding some missing functions: Flip/articulated screen, In Body Stabilisation System and audio output jack. (Why SIGMA way)?, Those are all given and can not be changed now. When it comes to functions that are already included in this camera but could/should function better, I would like to highlight the following:

  • Flickering image can be very obvious in all recording formats and frame rates. Currently the only way to combat this phenomenon is by applying a de-Flicker filter on the footage. (I had to De-Flicker many of my shots in the above video).
  • The camera is using a contrast base focusing system and unfortunately, it is struggling to maintain constant autofocus performance when tracking moving objects.
  • Electronic image stabilization can be better than having nothing but in our case, it can be activated ONLY when working in MOV filming format. Apparently there is a technical challenge to apply it when filming in CinemaDNG maybe due to the crop factor needed when deploying it.
  • I’ve already mentioned my take on the MOV picture quality. I wonder if anything can be done via firmware update in order to make it more capable.
  • The camera has a few picture profiles (let’s name them “Color Modes”), but the essential flat LOG picture profile is missing.
  • If the menu structure can be improved it will be a treat. In addition, more flexibility to assign functions to the camera buttons is needed.
  • There is a phenomenon that when pressing the shutter button halfway for achieving focus in autofocus mode, the LCD screen is becoming momentarily darker. This is a small thing, but one that greatly disturbs throwout the working day.
  • If you are planning shooting anamorphic, take into account that currently, no “de-squeeze option is available. It can be that there is a technical limitation in implementing such function when filming in exposure modes other than “manual”, but as we all shooting with fixed shutter speed and given ISO at a time, maybe this function can be considered and implemented. 
  • 4K/60 is becoming a broadcast standard and it would have been nice to see SIGMA supporting this frame rate in 4K resolution at least in MOV recording. 
  • When filming in CinemaDNG 12bit, only a single frame rate is supported (23.97 fps). To our European friends, you are restricted to choose between 8 or 10bit recordings when filming in 25 frames per second.
  • Previewing footage from within the camera when filming in CinemaDNG (Or deleting a file if not needed) is currently impossible. According to SIGMA, the possibility of previewing video will be added at a later stage via a firmware update.
  • When filming in CinemaDNG RAW, I would have loved to see the possibility of adding a metadata flag to the footage so when imported it to DaVinci Resolve, one will have the possibility to mute the recorded color and start working on a flat image profile directly. 
  • The camera can not be operated/charged via an external battery when switched on. Charging via an external battery is possible only when the camera is turned off.
  • Speaking of charging, a charger is not included in the camera package. Take this into account when planning your shooting day as the camera is your charger and obviously can not be used while charging.
  • Using “LCD touch screen to autofocus” is possible but when placing the original Viewfinder on top of the LCD screen, this functionality is of course not possible…Currently, the back wheel keys do not support focus point placement when recording.
  • Magnification while recording for refocusing is not possible. The magnification function will work only in standby mode. 
  • The original SIGMA VF is is a nice well-built product but a bit limited in its diopter strength. (+1 to -2). Take it into account if you are wearing glasses or contact lenses.
  • No audio output. Monitoring the sound coming out of the camera is not possible.
  • MOV format dynamic range is limited for 8 stops only!

Additional note:

SIGMA fp – frame/bit rates

  • The current maximum recording time is up to 2 hours.

SIGMA fp Camera lab Test – Rolling Shutter & Dynamic Range

Gunther, the man behind our lab tests has put the SIGMA fp camera through its’ paces and here is what he had to say about the camera. (If you are not familiar how we test dynamic range than please refer to https://www.cined.com/the-cinema5d-camera-lab-is-back-dynamic-range-tests/)

Running the SIGMA fp through our cinema5D lab to test the Dynamic Range actually left me with more questions than answers.

I was very much looking forward to this camera – a small full frame filming device with RAW recording capabilities and the right mount to use all my glass – As far as I’m concerned, a dream come true! I was close to pull the trigger and get this cam but for now, let’s take a closer look at what it can do. 

Looking at the lab results it is a mixed bag. Here is why:

  • There is no official documentation from Sigma’s side on the color & gamma settings for the cinemaDNG sequences. And most importantly, a flat or LOG picture profile is missing.
  • Strangely, the noise floor is partially cut off by the way how the code values are distributed. For the MOV ALL-I H264 4:2:0 8bit files, IMATEST could not calculate the noise values in our signal to noise ratio of 2 threshold criterion – those lower stops simply don’t exist.
  • The CinemaDNG footage was flickering significantly, therefore results were not fully consistent between the individual frames.

So, step by step. Let’s start with the good results: 

Rolling Shutter measurement

As mentioned earlier we are filming a strobe light at 300Hz which leads to a series of black and grey bars due to the nature of readout of CMOS sensors. Each pair equals to 3.3 [ms] rolling shutter.

In the rolling shutter department, the SIGMA fp fares quite well – at full frame 3840×2160 23.98, 25 and 29.98 fps settings, I am getting 20.8 [ms] of rolling shutter.

This is a very good result for a full frame sensor. The Sony a7 series have a rolling shutter of about 25 [ms], the Panasonic S1 has a value of  22[ms] for comparison.

SIGMA fp – Rolling Shutter Full frame UHD 23_98p 20_8ms

 

Rolling shutter halves at the full frame 1920×1080 settings for 23.98, 25 and 59.94 fps: 10.4 [ms].  This is a very very good value, but unfortunately a clear sign of skipping lines during readout of the sensor.

Ideally a proper downscaling of the 6K sensor is used to get a really good 1920×1080 image – from the rolling shutter measurement, we can conclude that this is not the case.

SIGMA fp – Full Frame FHD 23_98p 10_4ms

Dynamic Range for 12bit cinemaDNG 3840×2160 ISO 100 (Native ISO)

This mode is only available via external recording. PremierePro (latest version) does not recognize CinemaDNG files hence I moved to DaVinci Resolve.

Here, we are left to speculate on the color space and gamma settings in the Camera Raw tab to develop the CinemaDNG sequence:

Resolve CinemaDNG settings

The picture profile we used, in this case, was NTR (neutral). There is no “flat“ or “log“ picture profile available – which is clearly limiting the potential of this camera. Let’s hope this will come in a future firmware update!

Have a look at figure 1 and figure 2 for waveforms of our Xyla 21 step chart at “sRGB“ and “2.6“ gamma curve settings.

For “sRGB“ the noise floor is almost nonexistent – all the lower stops are buried. For „2.6“ gamma curve settings, something similar to a noise floor becomes visible, but the lower stops are very faint and almost nonexistent.

For gamma 2.6 settings the Dynamic Range reading of IMATEST shows about 11.3 stops at a signal to noise ratio of 2, see figure 3, which is also evident from the waveform at figure 2. But this result varies a bit from frame to frame as the image flickers – almost like filming neon lights (regardless of what you shoot). I have no explanation for this flickering.

The lack of a noise floor also means that you will have difficulties protecting your highlights and still getting meaningful information from the shadows at the same time. 

Fig. 1: Waveform plot of the Xyla 21 step chart using sRGB as the gamma setting for CinemaDNG 12 bit UHD at 23.98fps ISO100. The 11th stop starts to vanish in the noise floor – which is almost nonexistent.

Fig. 2: Waveform plot of the Xyla 21 step chart using a gamma 2.6 setting for cDNG 12 bit UHD at 23.98fps ISO 100. The 11th stop is clearly visible, but the 12th starts to vanish in this faint noise line above the 0 code value line.

 

Fig. 3: Dynamic range result from IMATEST for UHD 23.98fps cDNG 12 bit ISO 100 at a gamma setting of 2.6 – 11.3 stops are shown for SNR = 2. However as is evident from the strange code value distribution (top graph of the 3 on the left side) the lower stops of the step chart are very faint and almost nonexistent. Strange.

Dynamic Range for 12bit cinemaDNG 3840×2160 ISO 4000

I was curious if a noise floor would be there for ISO 4000. Looking at the waveform plot for gamma 2.6 settings below (figure 4), yes, there is a significant noise floor now. And the dynamic range result drops to about 9 stops. IMATEST reads 9.25 stops at SNR = 2.

Fig. 4: Waveform plot of the Xyla 21 step chart using a gamma 2.6 setting for cDNG 12 bit UHD at 23.98fps ISO 4000. 8 – 9 stops are visible above the noise floor, which is very pronounced.

Dynamic Range for H264 8 bit 4:2:0 MOV All-I 3840×2160 ISO 100

The first surprise for the MOV H264 recording – a very low bitrate of roughly 28 Mbit/s is encoded.

Next, the noise floor is completely cut off in this mode, see the waveform (figure 5) below – 8 stops are visible before the zero code value line is crossed. IMATEST, therefore, cannot calculate the noise for the lower stops – nothing is there.

Fig. 5: Waveform plot of the Xyla21 stepchart for internal UHD 23.98fps H264 8 bit 4:2:0 All-I recording at ISO100. The 8th stop barely makes it above the zero cut off line – there is no noise floor.

Lab Test Conclusion:

The lab test results show a mixed bag. The rolling shutter result of 20.8 [ms] for a full frame sensor at 3840×2160 readout is very good. The dynamic range results – despite having the capability of RAW recording in CinemaDNG format – leave room for improvement. I think this camera has not reached its full potential yet. Future firmware updates could certainly change the game here – for example, if a LOG picture profile would be implemented.

Conclusion

The fp camera is SIGMA’s first attempt to enter the saturated market of mirrorless cameras that shoot videos. In my opinion, its uniqueness lays in its compactness, mobility, sensor size and the ability to capture images in RAW file format. In regards to mobility, this camera screams “take me out and start shooting”. I hope that SIGMA will consider favorably developing a line of compact stabilized L-Mount lenses (Zoom or fixed focal lengths) to go alongside with it.

If you ask yourself who is this camera for, well, it is not easy to answer. Uncompressed RAW is great for certain filming scenarios but for “every day (documentary) work” it is a REAL pain. For greater flexibility, I hope that better implementation of the MOV format can be considered.

When it comes to Dynamic Range measuring vs. Images, well, charts are charts (And we should not ignore them), but at the end of the day what counts is the actual image that comes out of the camera. I’ll let you judge by yourself if you like what you see or not…

I’m also happy to report that after communicating directly with the good folks at SIGMA, I heard that they are fully committed to enhancing the functionality of this camera by future firmware updates whenever possible. 

EDIT: SIGMA acknowledged our findings and released an official statement. We are happy to see they are taking care of it and working hard to solve the issues. 

 

Man at Work with the SIGMA fp camera (Image credit: Shari Lai)

With Mai Watanabe-san. Image Credit: Sheri Lai

Many thanks to Mai Watanabe-san for contributing with your precious time. It was truly a pleasure spending half a day with her!

The above video was shot on a SIGMA fp camera. (Final hardware but not a final firmware so final images can be different). Picture mode: NTR, 23.97 fps. Recording format: Mostly CinemaDNG 12bit. (Mai-san performance was recorded in CinemaDNG 8bit) and I sneaked 2 MOV shots too…Edited on DaVinci Resolve 16 latest beta version. Unfortunately, Adobe Premiere does not fully support its own RAW format files for video editing. Lenses used: SIGMA 24mm and 45mm L-mount and SIGMA 24-70 Canon EF mount. Colored with Lutify.me.  I used tripod for 95% of the shots. The rest handheld  and  some stabilised in post. All under available lighting conditions.

Music courtesy of MusicVine.com – Get 25% off any Pay-Per-Use license with code C5D25 (valid for one use per customer).

You can find out more about the SIGMA fp camera by visiting the official SIGMA site.

68
Leave a Reply

guest
1
1
- 3
Filter:
all
Sort by:
latest
 Nishant Gala
Nishant Gala
Member
November 12th, 2019

I used the camera in its most basic structure and simple shoot to see how it hold up in challenging and harsh scenarios. here is the link to the video. do chk out and experience urself. also do like share and comment . also uploading soon the cinemadng 10bit and 12 bit footage test of the camera. CLICK or copy paste url
https://youtu.be/4xPyzcoTPEY

Brian Williams
Brian Williams
Guest
November 8th, 2019

New firmware just got posted for FP, fixes the external recording issues with SSD/flickering, pretty awesome.

antoine amanieux
antoine amanieux
Guest
October 29th, 2019

how can a FF sensor have less DR than the bmpcc4k’s m43 sensor ?

Member
October 25th, 2019

First I appreciate the work you’ve put into this. A lot of useful information for sure. Just wanted to share some feedback.

I’m just not a big fan of these reviews when “image quality you see above is NOT final as I was filming with a NON final firmware”. Especially when there are lab tests etc.

Is this the camera manufacturers that insist on getting reviews out fast? I understand that you as a news site want to bring out news asap. However, I doubt you will do as extensive tests when the final version of the camera comes out. I certainly would not have the energy to do it all again! :D

The reason I don’t like it is because in maybe 3-6 months time, when I want your opinion about a certain camera, I will find these types of reviews that have tremendous amounts of work put into, but the conclusions are based on wonky firmware that may or may not differ from the final version. Maybe the noise-floor stuff are not even there in the final version etc.

If it indeed is the camera manufacturer that want you to put out the review asap and don’t care if the firmware is final, I think it’s bad strategy from their side. Especially if the image quality will be better when the final firmware comes out. Because the footage and tests in this article is the only footage they will be judged by (by me at least, initial impressions matter a lot). If they want to go ahead with these early firmwares, then I think they should have the “balls” to say that image quality is more or less final so that you can state that in the article.

If it is more your decision, I would opt for longer waiting time for us, so that you can get the final camera and do lab tests with the final firmware. Then the article itself will live longer and you won’t need to do a follow up article.

That said, thanks again for all your work and valuable opinions Johnnie!

Member
October 28th, 2019
Reply to  Johnnie Behiri

Aaah I see. I understand why you wouldn’t abort the filming just because the firmware wasn’t ready when you’d come all that way. I didn’t understand that the lab-tests were indeed made on a final firmware, it wasn’t so clear in the article when I read it.

Glad that we could have a dialogue about it, looking forward to more articles from the trip!

 Dennis Schmitz
Dennis Schmitz
Member
October 22nd, 2019

The DNG samples provided here look amazing to me.
Great detail, fantastic texture (something I miss on BMD cameras since Braw and on the S1H because of temporal NR…), great colors decoded as BMD Film and even the 8 bit sample shows great dynamic range and seems to be free of banding artifacts.

Marc B.
Marc B.
Guest
October 22nd, 2019
Reply to  Dennis Schmitz

Your input is of particular relevance to me. I’ve held off on the BMPCC4K & Panasonic S1H until I can evaluate the SIGMA fp. I much prefer the CinemaDNG RAW to the BRAW.

A
A
Member
October 21st, 2019

So the battery life is not that bad even using and external SSD? Did you expend more than a day filming the short documentary? If so how many batteries did you need it for a normal shooting day? I would say for a Panasonic S1 I need 2-3 batteries, with the X-T3 3-4 and for the GH5 2-3 batteries.

A
A
Member
October 21st, 2019
Reply to  Johnnie Behiri

Thanks for you reply. Seems OK in pair with the X-T3. Hopefully FW improves a bit more

Chris
Chris
Guest
October 19th, 2019

How would log improve on the DR of images shot in linear RAW?

 Nikolas Moldenhauer
Nikolas Moldenhauer
Member
October 19th, 2019

I watched the video…. and I didn’t think about the camera for a second. Good job Johnnie. Captivating pice about Japan.

 Paul Curtis
Member
October 19th, 2019

Johnnie, one final clarification (i say final but it probably isn’t). You have confirmed that the UHD is full frame – so the full sensor is 6064 x 4042, cropped to 6000×4000 (RAW digger shows all the pixels). So that’s 3:2. The UHD is 3856×2170 (cropped to 3840×2160 no doubt). From a FOV view is that a crop of the centre of the sensor – so the vertical fov is the same between stills and video?

I just wonder what they are doing 4042->2170 does not really scale well. But then a centre crop would make for a large fov difference…

Did you ever get a chance to lock it off, shoot a still then shoot the same thing in different DNG versions – then we have a way of confirming the quality differences between stills and movies and then also different depths of DNG…?

One positive about DNGs and low depths is firstly the depths are pre matrixing and white balance, so even if the RAW channels are 8 or 10 bit by the time they end up as colour then the effective tonality is much higher, less chance of banding – compare to say a baked movie format of 8bit H264.

cheers
Paul

 Paul Curtis
Member
October 19th, 2019
Reply to  Paul Curtis

I’d also add that after some more digging the 8 bit DNG includes a linearisation table, so it seems that sigma have done the right thing and made the 8bit ones non linear, which is great news. The 12 bit appears to be linear though.

 Paul Curtis
Member
October 23rd, 2019
Reply to  Johnnie Behiri

Thanks Johnnie, that’s awesome news. If you get to shoot again then can you do a 10bit version. If the 10 bit is not linear (like the 8 bit) then you would never really need 12 bit linear. I really hope that’s the case because then you have all the frame ranges for PAL at 25p.

(Sorry i just read that you said 16:9 stills mode)… Surely the movie though is cropped a bit? The sensor is 3:2 native but we’re extracting 1.77? So the fov can’t be the same as stills?

thanks
Paul

sam broggs
sam broggs
Guest
October 19th, 2019

“I am here to meet my son”. Where is the family reunion? :)

Vlad Box Rojas
Guest
October 19th, 2019

Love it. I remember staying at a Ryokan in the Fukui prefecture, and meeting the Geishas. I also recall how extraordinarily expensive were the Kimonos. Great piece, I think the camera looks great very natural and close to the colors of the Japan I remember.

Oscar M
Guest
October 18th, 2019

Great review as always –
– I was really surprised with the bad colornoise in the night cab shots around 04:30
even my tiny sensor bmpcc4k looks way better in low light?

Scott
Scott
Guest
October 19th, 2019
Reply to  Johnnie Behiri

Hi Johnnie – thanks for the thorough review and absolutely excellent short film. That shot you mention is my favourite… the lighting and look is very cinematic. Even the rounded windows in the taxi add a retro feel to it.Bravo!

Oscar M
Guest
October 19th, 2019
Reply to  Johnnie Behiri

Ah, OK Johnnie that makes sense – looking forward to test one myself now that I have my leica lenses for the TL2.

 Richard James Milne
Richard James Milne
Member
October 18th, 2019

Wow, extensive review! Great story!! Sigma has ways to go… but if they ever make a camera like there ART series lenses, I’ll be buying one :)

JR Lipartito
JR Lipartito
Guest
October 18th, 2019

Great story, I hardly paid attention to the image quality—which is a good thing! The fp can clearly do the job it’s meant to when in capable hands. I’d be interested to see how well it can cut together with the S1H. In form factor alone these seem two very complimentary cameras and it’s a big plus that they share the same mount.

 Wayne S
Wayne S
Member
October 18th, 2019

Now for the bad side. This was bound to be a bit of a bummer to some degree with a foveon coming and at the price range. No 6kp50, no 4kp50, no 14 bit, in that order. No charging, or bypass external power, while in use. No preview cdng in camera, or meta data issues. The 8 stop non cdng. It has to be compared to other small competitors. I know people genuinely disagree to give a new camera benefit of the doubt, but at this price, everything said here can be added. Do you really think this costs $1600 more to make than a $299 micro four thirds camera (Yi-M1 on special, which has interesting (still) pictures which reminds me of this camera).

If they could have had 12 bit 4kp50, or 12/10 bit 6k/p50, or foveon, I would be tempted to buy this. Now, to wait to see the next micro and mini lines next year, and foveon camera, hopefully with 1:1 RGB per pixel.

HDMI carries audio, if there was an adaptor for that, you could monitor.

Brian
Brian
Guest
October 18th, 2019
Reply to  Wayne S

Weird, you just compared this to YI-Mi.

 Wayne S
Wayne S
Member
October 19th, 2019
Reply to  Brian

Not really, I just compared parts costs and look on an older Sony sensor. It’s really not that much more costs in it, value just maybe, but costs know. The YI-M1 is a true low-cost competitor for cost comparison objectively. You can’t use an expensive camera for comparison, as they are just extra markup to cover profit, inefficiencies and wages, the reverse of objective. Which would be weird.

 Wayne S
Wayne S
Member
October 18th, 2019

Nice footage Johnnie. I forgot I had the brightness on the IPS screen of this phone turned down, and it came out very well, reminding me if that Canon pocket hd1 camera years ago. However, moving the brightness down in shooting (is this just dual gain?) would have moved the song sheets in the table, her shirt, the light fixtures and the burnt out spot behind her performance, into a good range while looking like what I saw here, which could have gone darker again. Beautifully smooth, and the subtle encroachment of the bokeh, is what inspired me years ago.

But let me ask you, the first bit, outside in the sunshine next to the water, and in the car, was that still the FP? I could say XYZ about the bright areas being overexposed, but as above, just moving them, with the image, down in brightness, is going make it very filmic. You have a good eye. This camera is like what people wanted the pocket 4k to be. That camera’s image just feels off.

But, what is really happening with the image. No tiles posted, but looking at the graphs, we can see some pronounced lower stops compared to the pocket. I’m interested in what the stop range looks like, and what is happening with this low level clipping you mentioned. I’m interested in a mode which gives you more lower stops into the noise floor, to at around with. At the moment I’m counting a wishful 15 levels that you could endeavour to use (similar to the pocket 4k), but suspect up to two more might be there in a rotten noisy mode (or am I reading that screenshot wrong?). Looking at the data rates, the cdng is in lossless territory. A shame about no 6k/p50 foveon or 4kp50.

A
A
Member
October 18th, 2019

Jezzz that’s about 1TB for 1 Hour of Footage.

BOUNCE
Member
October 18th, 2019

It would have been good to know more about battery life. Especially enlight of the fact that the camera cannot charge while being used.

The flicker is a real disappointment. Do you know if Sigma plans to address this issue?

Mikhail
Mikhail
Guest
October 19th, 2019
Reply to  Johnnie Behiri

Hi Jhonnie! very cool review .what do you think the lack of charging while using can it be changed through firmware update (theoretically) or it is hardware restriction?

Daniel Rozsnyo
Daniel Rozsnyo
Guest
October 18th, 2019

So this is the first camera that not just fixes bad pixels in the pre-baked RAW, but also scales it to 2/3 of resolution? There is no way a 6K sensor can produce a 4K RAW file directly. The question is – whether that scaling comes from a 6k3k source, or a line-skipped 6k x 1.5k source, that gets re-interpolated into the 4k2k file. The scaling might be also the source of confusion for the imatest suite.

 Paul Curtis
Member
October 18th, 2019
Reply to  Daniel Rozsnyo

Im keen to get to the bottom of this. I’d assumed it was a UHD crop of the 6K sensor. I’ve heard conflicting reports about whether the DNG is full frame or not. I don’t see how they get 6K into 4K unless they’re using a really odd binning system and i don’t see how this is RAW if there is any sampling going on (I suppose you could resample 6K to 4K bayer channel by channel then create a DNG from that… In fact if you did that then you’d perhaps clean up some aliasing so maybe that’s it. Scale RGGB each separately and resample 4K from that…)

 Wayne S
Wayne S
Member
October 18th, 2019
Reply to  Paul Curtis

You are right Paul. Sounds a bit insane, but I have proposed rescaling and reconstruction of a misaligned Bayer image in the past.

 Paul Curtis
Member
October 18th, 2019

You should look at the DNGs in something like RAW Digger which will show you the data held in there – you can tell then whether it’s linear/log and the real bit ranges. Is there any chance of posting a DNG or two from the footage to look at? That would be awesome.

One thing that isn’t clear is that whether the movie modes are cropped or not. Is the UHD version a 4K crop from a 6K sensor or is it full frame? If it’s binning from a 6K sensor *and* no OLPF then that could cause troubles as well. It sounds like the 1080 is not a crop though but skipping or some kind of binning? Looking at the source DNGs would be really useful to understand differences between 8bit and 12bit for example.

For example if the DNG is 8 bit and the data is stored Linear then you have 255 levels but RAW also includes a black level and on the stills this is around 1000 (out of a 14 bit source). If that same black level is in the 8 bit and it’s linear then you’re looking at very few code values to store that range. Plus over half of those values would be used to hold the brightest stop of data…. So 8 bit linear in RAW IMHO would be useless – so i am hoping they have some encoding scheme in there to distribute the values differently – your observations about missing the bottom stops might just be an issue to do with linear source.

Again looking at the DNGs would help explain this…

cheers
Paul

 Lars Steenhoff
Member
October 18th, 2019
Reply to  Johnnie Behiri

Both files have great dynamic range even the 8 bit one.

I open them in Adobe Camera Raw with Photoshop. you can really move the sliders all the way and no problems at all.

 Paul Curtis
Member
October 18th, 2019
Reply to  Johnnie Behiri

So 8 bit does have black level subtraction so code values are from around 35 to 255, they do appear to be linear so IMHO 8 bit really isn’t going to be a lot of use, especially in darker environments. Linear encoding means that light doubles in value – so outside if you expose as far right as possible you can maximise the small container. In the water shot there isn’t really a lot of moire on the water which is a good sign.

The 12 bit is tonally a lot nicer, black levels are at 256 but in a 12 bit container you can go up to 4096. The shot of her has no highlights in so there are a couple of stops above the brightest point empty. Again i don’t see horrific moire in the patterning of her dress either.

So in the shot with that wire window is there a workflow issue? Like a screen grab from a preview quality debayer?

There’s always something so much nicer with no compression in and even though i’d say the imagery is quite soft, it’s a nice soft. So DNG is the way to go.

It’s a real shame if they cannot push 12 bit UHD to 25p. The datarate is there because 29.98 at 10 bit is higher.

But i am really interested in finding out whether this is a UHD crop of a 6K sensor or whether there is something else going on?

cheers
Paul

 Wayne S
Wayne S
Member
October 18th, 2019
Reply to  Paul Curtis

I’m interested in where the missing 6k goes too. A 4k crop is going be close to s35 sized, rather negating the user of full frame format for filming. But I’m glad something is here which gives full frame depth of field adaptor like images.

 Dennis Schmitz
Dennis Schmitz
Member
October 22nd, 2019
Reply to  Paul Curtis

Could it be some 3x binning?

 Lars Steenhoff
Member
October 21st, 2019
Reply to  Johnnie Behiri

Do you have a frame of 1080p cdng?
Would love to check it out

 Dennis Schmitz
Dennis Schmitz
Member
October 22nd, 2019
Reply to  Johnnie Behiri

These samples look fantastic to me, decoded to BMD Film and my own Lut applied.

Beautiful color and incredible amounts of texture.

 Dennis Schmitz
Dennis Schmitz
Member
October 23rd, 2019
Reply to  Johnnie Behiri

Please upload the full sequence if possible
https://drive.google.com/open?id=1M1ye3WadZJk4dAWf0PRGFQ3WpL63yQiy

I love what I can get out of these DNG files, even the 8 bit LOG cDNG file looks great. Want to try out some different deflickering techniques.

 Jordi Mlro Mlro
Jordi Mlro Mlro
Member
November 14th, 2019
Reply to  Dennis Schmitz

I went somewhat in the opposite direction in grading, but yeah, the images look pretty fantastic:

https://drive.google.com/file/d/15JKoZvDEHXawGzIlXntzxe5pm81N7Mro/view?usp=sharing

 Wayne S
Wayne S
Member
October 18th, 2019
Reply to  Paul Curtis

+1

 Lars Steenhoff
Member
October 18th, 2019

And for developing cdng in Resolve, could you not choose black magic film instead of rec 709 to get a flatter image?

 Lars Steenhoff
Member
October 18th, 2019

One question I have not seen answered yet, is CDNG possible in full frame and in 1.5 crop mode?
Or is it limited to one mode?

 Lars Steenhoff
Member
October 18th, 2019
Reply to  Johnnie Behiri

Thanks that’s great to hear!

 Jones Paul
Jones Paul
Member
October 19th, 2019
Reply to  Johnnie Behiri

Thanks for the review Johnnie. I am a little confused and need some clarification- will this camera shoot CDNG “full frame” in both 4k and 1080p? As well as 60fps 1080p full frame? This spec will sell me on this camera as a usable b-cam, stealth cam, BMP4K replacement. Cheers Paul

 Lars Steenhoff
Member
October 18th, 2019

Looks really good!

It does seem that there is no antialiasing filter on the sensor, as some moire can be spotted at 2:04
https://youtu.be/qQTiJiDt0y0?t=124

But this can be solved with some filter on the lens

JR Lipartito
JR Lipartito
Guest
October 18th, 2019
Reply to  Johnnie Behiri

It would be great if a third party like STC would make an insert OLPF. You’d never have to take it out since there’s no mechanical shutter in the way.

1
1
- 3
Filter:
all
Sort by:
latest

Take part in the CineD community experience