Kinefinity MAVO Edge 8K | Dynamic Range

iiiNelson

What are the dynamic range figures on the Mavo Edge and how do they compare with the Mavo LF? I’m really torn between both cameras to be honest (or if I will even end up adding a cinema camera to my kit). Both look great with the Edge having quite a bit more refinement it seems. I do like the LF look still though as well and I really hope to see Kinefinity move into the US/North American market in a major way.

I know there are smaller sensor sizes that are more affordable but I’m only ever going to consider LF video options as those are the lenses that I own without having to worry about crops.

Jerome

iiiNelson
Before stating simple numbers we first need to know we are having the same definition of dynamic range. Here at Gafpa Gear we work with different definitions. Here at Gafpa Gear we state that by far the highest scoring camera in the Latitude/dR game is the Arri Alexa, it scores 13.5 stops of dr in pixel to pixel and can be upgraded to 14 stops oversampled by 70% to 2k. By oversampling the latitude doesnt really change that much just a bit cleaner in the shadows and a bit less color pollution by lower chrome artefacts.
Dynamic range is often calculated with a IMA test, you simply show the camera with a neutral lens with a good contrast reproduction, a few patches of light, every patch represents a stop of light. We call these charts XYLA charts, a back illuminated chart with patches with filtration filters, that will get you absolutely scientifically correct stops of light, without color polution, etc. Look at them as ND filters.

Once you ve shot a Xyla chart you can simply see what amount of stops above the ADC noise a camera will capture before it clips. CLipping in the highlights is a brickwall, while clipping in the shadows is a bit more of a topic where people start to dream and fantasize and turn their alter egos into harsh populist with fake news concepts. We can talk hours about this, but basically a 12 bit adc will never give you more than 12 stops of DR. The arri Alexa uses a very high end sensor, which simultanously has a ADC set for different gain levels, with a crosstalk and can combine these values into information that translates close to 14 stops of DR. Not only is the DR very good on such a camera also the color response is pretty linear, which means that if you would underexpose by a few stops, you can push back the image to normal without strange color shifts or color artefacts, ofc the more you underexpose the more noise you will get because the more you get to the limits of a sensor and the resolution of the adc the higher the standard deviation and the more noise that will occur, once we get in the adc noise which is basically dark current not part of the visual image, we might be able to retrieve some sort of visual information (overspill), but we are basically bottlenecking. By executing Noise reduction in camera you can fool a ima test. an ima test is based on the believe that a camera doesnt have internal noise reduction, it sets a threshold for noise often a value which is close to its adc noise (dark current), which of course varies from camera to camera, you can fool a ima test a bit by oversampling (quantization of pixel averaging them out, for lower noise values) , but you cant fool a latitude test. A latitude test is simply a given scene which you will shoot in 6 stops under and 6 stops over, often you put some fine detail in there like a human (with hair), and skintones, a color passport, some specular highlights, some deep shadows. Often the combined dr of such a scene is spot metered around 8-10 stops, so any camera can capture it when exposed well. Once you start to over expose and underexpose you can see which camera can be corrected in post to that normal exposure without clipping, too much noise, or lost of contrast and colour shifts and artefacts like posterization, fpn (fixed pattern noise), and fake highlight stops, or highlight stops that clip on different levels in different colors (alexa clips at first in the yellow channel).

Even though a shot Xyla chart will still show some shadow patches this is to be considered ‘ macro contrast’ , wich means that the camera can only detect details on a Macro level which translates to a shade of the curved roof underneath a bridge for instance.

But because the noise is _almost randomized (completely randomized means that its unable to compile any visual information), it cant display fine detail, like a hair. You can make details less randomized by denoising or oversampling, both will kill spatial resolution thus lowering the resolution and the ability to see fine detail, so whether you denoise or dont denoise it wont really mystically show more information, all that happens is that noise values are less aggressive/extreme. The only way to get more data out of these shadows would be a AI (artificial Intelligence) algorithm that extrapolates info that doesn’t exist by recognizing shapes and filling gaps.
So long story short you might be able to see some patches on a Xyla chart on the deep shadows, but that doesnt mean that they should be considered to be a part of the latitude (actual dynamic range), by fooling a ima test with Noise reduction (Sony cameras), or fake Highlightstops( Red), you only get a better numeric value out of your ima dr test, but this doesnt translate to a better image in real life.

Also Ima test has proven to be able to fooled even worse by rating a camera beyond it’s ADC capabilities, another reason to do Latitude tests.
Latitude test are easy to understand, both by adults, Childs, populist presidents, rapist, psychopaths and dolphins. They simply show how many stops are usable, and that’s what we should call the dynamic range. The downside is that it’s subjective what we consider to be ‘ usable’ and you should execute the same latitude test with each camera so you can line them up with their clipping points so one can see which camera performs better. Thats the fun part, most often people talk about which camera clips faster… But you should turn the concept upside down, a sensor has a well capacity, if it gathers more electrons it clips, but on the other side of the spectrum we have the noise floor , and the adc resolution. The better the sensor and its ADC (adc is not the only thing that determines how much dr you can capture!!) the more stops you can capture in the shadows. Imagine a good sensor with good signal to noise specs, and it has a 12 bit ADC mode, you can now potentially capture 12 stops of light which can be used as usable stops, once you change the readout mode to 14 bit ADC you will get theoretically two extra stops in the shadows, but the image will clip at the same point (well capacity).

Pixel Saturation
Saturation is a phenomenon which occurs when an individual pixel is no longer able to store any more charge (i.e. cannot accept any more photoelectrons). If each pixel can be assumed to be a well of electrons, saturation occurs when this well is full.

The maximum charge that can be stored within a pixel is largely determined by its area, however as the charge is stored within a potential well, the probability of trapping an electron within the well reduces as it approaches saturation.

Therefore, the linear relationship between light intensity and signal degrades as the pixel approaches the saturation limit. This results in an apparent responsivity of a “saturated” pixel drop. Well capacity is sometimes referred to as the linear full well. This is the point at which the linearity between light intensity and signal degradation exceeds an acceptable level, not when the pixel is at its ultimate saturation limit. Cameras are usually designed so that this signal level fills the full dynamic range of the analog-to-digital converter.

Before we will continue with the more practical side of this small essay and finally answer your question, I have to admit that I only partly understand these technicalities, and even though I dare to point them out I might be wrong in some tiny details, and If so please point it out to me, so I can update this post, so the quality will be better and better, and can eventually serve as a school post on our GAFPA GEAR website.

So in the next post I go in depth with my findings of the Kinefinity MAVO LF and Edge.

https://gafpagear.com/shop/kinefinity-mavo-edge-8k/

Jerome

iiiNelson
SO now that I kind of layed out the definition of dynamic range and the technicalities involved, we can look more closely at Kinefinity cameras.
For most newbees Dynamic Range ins a bit struggle, and they often feel that their work improves a lot if they switch to a high DR camera.
This is ofc the reason that most camera brands (except ARRI), fail to deliver the truth on their spec sheets, it’s even worse, it’s sooo spoiled that you can’t survive as a small brand to state honest DR numbers. Therefor I think we should keep talking about dr, not why you need more, but what it is!. If we look at most high end feature films, they are mostly lit within a 10 stop range and mastered to be displayed within a 6-8 stop range (in cinema you have quite some global light, and you probably wont be able to see more than 8 stops). You can capture 16 stops of lights, and then retain everything of these stops and display these stops in a 8 stops display, but every two real life stop of lights will be clamped inside a 1 stop light step, which eventually makes them dull and not recognizable, and your image will lack power.
It’s a cliche to say that you dont need 14 stops, because a display cant show 8 stops, , some dops do a lot of ‘ re-lighting’ in postt by means of power masks (revenant is a good example, shot in available light) in order to do so everything has to be retained. But generally looking at cinema history the craft mostly is about getting your mis en scene within a 8-10 stops range, and only use the last few stops for a roll off, to make the highlights look less clipped (roll off means gradually rolling of to pure white). To my believe a good digital camera should be able to capture 12 usable stop, and a cinematographer and AC needs the right tools and knowledge to nail the exposure inside the camera (mapping the scene balanced over the available camera stops, without clipping or underexposing). the rest of stops on like a Arri alexa are not really there for creative manners, but to make you feel more confident on a shoot, because you dont have to set your exposure hundred percent correct in-camera, and you have a bit of wiggle room. FOr these reasons I (Gafpa Gear) developd ETTL luts, which help the dop, provide luts to distribute the brightness of a scene in the best way, read more about it here (under lut)

https://gafpagear.com/kinefinity-support/

-

-
The Mavo Lf uses a IMX sony sensor initial designed for photography, in photography you dont need continues shooting (lets say 25fps), and you rolling shutter is not an issue (either solved by electronic shutter or analogue shutter), so when using these same sensors inside a photography camera and making stills you can acces higher ADC bit readout modes (adc stands for analogue to digital converter), so these sensors in stills mode can all of a sudden resolve around 13-14 stops of lights. thats the reason that these cameras once entering movie mode are all of a sudden lacking dynamic range, even though there are log curves that help stalling the full dynamic range in a efficient manner, they are simply limited in DR by their ADC mode, because once shooting continuous frames the adc is not capable to have a resolution over 11-12 bit . It simply next to impossible to design a 14 bit adc that can do fast readout, hence the long readout time in stills mode which is not able to be fast enough for faster than a few frames per second. Some brands came up with ADG, dual gain readout, it doubles the readout time, thus you cannot shoot high fps, and the rolling shutter is worse, it first shoots one image at high gain, and then another at low gain, these values are combined, leading to higher dynamic range (by pushing the iso you can catch more stops out of the shadows with the same adc,) .

The ARRI Alexa is one of the only cameras which can simultaneously readout with two different gains. hence it can do almost artefact free fast readout, which both diminishes rolling shutter artefacts and let you do high fps.
BAE/Fairchild Imaging sensors used in the BMCC & OG Pocket and the 4.6K sensor are also true Dual Gain Architecture but only 2×11-bit, which doesnt lead to much higher dr numbers than regular 12 bit adc cameras. Another options (the least fun one), is HDRX (probably patented by them for the next trillion years, including their beautiful invention RAW etc) , its the least good one, it basically shoots one fram with normal shutter time, and then captures another extra frame with fast shutter, it combines these frames in post, leading to ghosting, and thus makes it unusable to use in anything else than a photo (once the image is moving you see the highlights ghosting as F**K) .

Dynamic range of Kinefinity camera’s
Kinefinity used for all their existing models except the EDGE 8k a sensor intended to be used with photocameras. Even though they can do video they are limited in terms of ADC (analogue to digital converter) bit resolution, if they would use the higher readout mode they might be able to do around 24 fps in some resolution but Rolling shutter would be so awful that it would be next to unusable.
Kinefinity is a relative young company, not having the money to develop their own sensor or get exclusivity on a fancy sensor, but they have quite a history of getting a lot out of their ‘ off the shelve sensors’ . On top of that they have flirted with filmmakers with their base Matrix which is reminiscent of Arri Log c and very easy to Grade with minimal corrections, for those who are after the so called ‘ film look’ which is a subtractive color system. of course sensors don’t work like a subtractive color system like film but their color matrix (color science ), can be tuned to behave like that. We call it the watercolors, because the brighter you go with a watercolor on paper the more white that shines through, and thus the lower the saturation, while additive (which is inherent to digital capturing with a traditional sensor) is the other way around, the brightest can also be the most saturated.

A sensor is only a minor part in the full image chain from lens to olpf, to cfa to sensor to ADC to NR to matrix to log curve to codec,, and thousand steps in between. And like I say if a camera can have the right settings and tools to give the DOP an idea of how to map the brightness of a scene in the most efficient manner (retaining highlights while having acceptable shadow details) , a dynamic range of 11-12 stops can already be quite nice. But like everything in life : more is better. I am a European style shooter, I love my colors and contrast dense, I like to retain highlights, I hate clipping them, so I am (especially doing documentary where i can barely control light) always fighting my shadows in order to retain enough highlights, I made a obsession about it, finding lenses that have somewhat lower contrast but with enough punch with only some lifting in the shadows, I experimented with Schneider Digicon and Tiffen Ultracon filter to lift shadows (lifting pedestals ), I even made my own varicon filter ( a illuminanted ultracon glass to locally lift contrast without milking the rest of the image), but eventually I knew these tools are ment to be used creativily and will lower the dynamic range of a scene, and not higher up the dynamic range of your camera, so its somewhat limited, because even though you dont clip, you also are not able to capture all details that the scene has to offer to you. I love the grades I did with the MAVO LF, i keep revisiting them, but I always look at the shadows, and know that they could have looked deeper, they lack that center spot which is deeper (lower in luminance, yes its close to what I want from my Look, and I know its a minor detail,

I basically would love to see the same exact image with the average pedestal of shadows slightly lifted like it is now, but some spots in the shadows to rate 1 stop deeper (i generally only use the first stop in my master grade as pedestal, to open up the image and dont like these kind on OLED curshed larger than life shadows, but i Love to see a tunnel, a infinity point of black, as if deeper black exisst in there, if you could travel there.

So yes after 2.5 year with the MAVO LF I think I got to know the look, and I could eventually set boundaries, knowing that in some parts I could improve but in other fields the camera should offer more to give me more of what I am longing to. In that regard I think any beginning dop should get a 8 stop DR camera, and make magic with it. Orson welles and citizen kane looks soft and deep, while all scenes are spot metered not to exceed generally 8 stops, thats the film look!. YEs its fun if your experienced to have more dr in your camera, but first you need to work with your eyes and to be able to feed the camera with a sexy looking image, and 99% of the shooters out there seems to fail in doing so (oh gosh iam so pretentious and snobbish).

I did a kind of secret ima test with the MAVO LF and AGDok did a latitude test, even though their model had some blackbalance issue, (thus more adc and scanning lines showed up) you could cleary see that the MAVO LF is not the best camera out there dynamic range wise (also not the 2nd best), what it does greatly is offering a nice study codec, a tall sensor, and a cinematography minded camera with lots of cinema resolutions and UI concept, swappable lens mounts, and a very nice matrix, and ofc 12 bit codecs with non destructive compression (prores), and recording to affordable media, etc etc. The MAVO LF also proofed to be a winner in low light, and retain most of its dynamic range when shooting at 5120 iso, It was also one of the first affordable Large format cameras with a cinema minded concept. But it’s surpassed in almost every way, namely the MAVO Edge 8k, which is now finalised ) a few units exist and once the ongoing chip shortage is gone we expect to ship a lot of these. In the beginning I myself believed that the MAVO Edge was a MAVO LF housed in a better thought out body, with industrial connections, e-nd, and dual image processing (clean out for your director, or DIT). making it a much more serious camera when image quality alone is not enough. But once I had the possibility to shoot with this camera back in MAY this year, i knew right away that this camera is also a huge improvement over the MAVO LF in terms of iq , both in color and dynamic relation, mtf, . So here we are.

image courtesy of GAFPA GEAR 😉

DYNAMIC RANGE NUMBERS
The MAVO LF should be rated somewhere around 11.1 stops of dr in 6k and somewhere around 11.4 in 4K oversampled from the full sensor readout,
This sounds very low maybe even unworkable but that’s how absurd the camera industry is, I have never encountered situations where the dr of the mavo lf wasnt suitable because its latitiude is nice, the stops of dr it provides ar good, (many cameras have somewha high dr but bad latitude_).we started to believe that about any camera can do 16 stops yet most of us (except Arri shooters) are used to cameras with a average latitude of 11 stops. We’ve been lied to, and only those who are willing to look around the corner will find out how wrong these ‘ marketing stops ’ are. I am not even going to say where and how they came to such numbers, because it would give them a excuse. it’s 100% intended to mislead anyone, including Donald Trump and Mark Rutte & Hugo de Jonge (look them up they are pretty annoying).

Kinefinity MAVO Edge 8K
I havent executed a IMA test with the MAVO Edge but since I have the data of the MAVO LF I was able to do a latitude test side by side (will post that later but first want to redo that test because it was shabby looking 🙂 (didnt effect the outcome but you know…for marketing maybe not the best to release).
The MAVO Edge uses a video dedicated sensor, and it seem to score about 1 stop higher in Dynamic range, and on top of that the shadows have less color shift, and less issues like scanning lines are not there, its just clean up until the noise floor. This is a superb result, because most of the camera’s out there have not more usable latitude than 11 stops, and some even believe (including me) that most cameras are 11 bit effectively hence the 11 stops outcome.
(shadow noise like scanning lines on previous Kinefinity cameras can be improved with certain post production workflows

https://gafpagear.com/black-balancing-kinefinity-cameras/ )

-

Also the CFA color arrays seems nicer, and due to a less dense olpf the camera is smoother (ou would say sharper, but i say smoother), why? follow this topic to go in depth with resolution.

https://gafpaforum.com/d/3-kinefinity-mavo-edge-8k-footage-vs-4k-over-sampled-footage/10

-

Asides from all other things that the EDGE does better( feature and image wise), we rate the Edge to be 0.85 stops higher in 8k readout than the MAVO LF in 6k readout, this is improved in 4k oversampled, but Again, lets not go there, oversampling will lower noise and reduce digital artefacts, but it doesn’t directly affect dynamic range, so you wont all of a sudden see more details, instead shadow details will look cleaner and less distorted, which kind of higher the usable amount of dynamic range you can deploy (depending on your taste and if you will use NR in post).
Long story short the edge will rate around 1.2-1.3 stops lower than a ARRI ALEXA (we lined them up in our own latitude test, and this is our subjective opinion about usable stops).

1.2 stops seems a lot but it isn’t; take your lens, try to clip your highlights, and now rock back and forth a stop under and a stop over, this is simply only a tiny bit, if you learn your camera well, and know how to distribute the brightness and knowing how to post produce, I am sure there wont be much situations in which the ALEXA would give you a better image. Matrix wise there’s a still a bit work to do because this is a completely knew sensor system, but we already feel at home with the matrix, and Kinefinity will release an even improved matrix in September , on top GAFPA GEAR will introduce a very technical and highly accurate MAVO Edge 8K to ARRI LOG C non destructive LUT. This is designed to transpose the matrix of the Kinefinity camera into LOG C, not really needed for my own workflow, but a lot of graders out there are conservative and the ARRI LOOK IS something to be religious about so every other matrix is
compared to the Arri matrix, and if it somehow is different it will be dismissed. A matrix is only a base, you can use your own grading techniques to transform the data in a way that it forms a better base for your regular grades. With out matrix conversion we give any dop whos daring to use the Kinefinity on his next project with a bigger team including a producer with moustache and colorist with sunglasses the ability to convince them into this 8k adventure. XXX JEROME
Hope this helps!

RandThompson

Thanks Jerome, looking forward to your further explanation of the Dynamic range of the Mavo Edge and Mavo LF!

iiiNelson

No that sounds great and personally I only care about “usable DR” and not manufacturer claims. I find just about everyone (as far as the major companies) except Arri to be largely dishonest in their DR figures since they tend to count noise as a stop even if there is limited to no data that could be manipulated out of it. A lot of people get upset that websites like CineD often rate cameras well below manufacturer claims I find their tests to be the most honest “popular” testing being done on the internet. Well that and they typically remain consistent.

Thanks for sharing your initial thoughts. Personally I feel that any camera these days is pretty much usable but it’s still nice to know where cameras stack up with one another… well specifically to Arri as I feel like most that they’re generally are seen as the “gold standard” on the video side and I still find Leica to be that on the photo side.

iiiNelson

Jerome

Thanks for sharing and that makes perfect sense. It is about what I expected when I looked at the image from cameras that they’d be in the 12-13 stop level of usable DR. It does look like the Edge has a bit more DR than the LF and some of this can be seen in some of the “flatter” looks without heavier grading. This may be what has me torn between the two. The LF seems to look more “finished but it could just be the LUT used. This isn’t a negative to me but a reality that more DR equates to a flatter starting point in most cameras which can mean more PP work in many cases if not using a reference LUT/preset. I’m pretty new to video (relatively speaking) but I’ve been doing serious photography for around 20 years or so. Much of my knowledge translates (composition and creativity wise) but I’m still learning to get familiar with NLE’s in general. Thankfully, I understand color and technical details and it’s true that I’ve worked with cameras on the photo side with a lot of DR. I always laugh internally when people scream about wanting “Canon colors and a lot of DR” because those two things are on opposite sides of the spectrum. Canon sensors famously have had some of the worst DR in the market… I mean their CMOS sensors were comparable to Leica CCD sensors in DR. This lead to a more finished look for many but I remember when the Sony A7RII was released and people complained about the “washed out look” initially because few were used to seeing a true DR of around 14+ stops of latitude on the photo side. I agree that 8-12 stops of useable range is what most people see as the sweet spot. My favorite photo camera look out of camera has been the Leica M9 (at base ISO 160) ever since it was released… to this day it’s probably the only digital camera I felt I ever loved.

RandThompson

Jerome,

It was like attending a Master Class on the DR range of large format Kinefinity Sensor Cameras. Thanks! I am really learning the Kinefinity philosophy for image reproduction.

Jerome

iiiNelson

iiiNelson It does look like the Edge has a bit more DR than the LF and some of this can be seen in some of the “flatter” looks without heavier grading. This may be what has me torn between the two. The LF seems to look more “finished but it could just be the LUT used. This isn’t a negative to me but a reality that more DR equates to a flatter starting point in most cameras which can mean more PP work in many cases if not using a reference LUT/preset.

Yes the MAVO Edge has about 1 stop more usable DR than the MAVO LF, but this doesn’t translate directly to a flatter look.
TO understand this principle we have to go back to real life and carry with us a spot meter, so we can see how many stop of brightness a normal daylight scene caries. If the amount of DR of a given scene doesn’t exceed the DR of a camera, and the camera operator, maps the brightness in the right way (thus no overexposure nor underexposure), the look will be as flat as any other camera that’s able to resolve the amount of contrast of the given scene. Overexposure and underexposure are most often user errors, and yes if you give someone a camera with a higher latitude/dr its less likely that he will make this error, because the camera provides more luxury (wiggle room) in that regard.

The next 4 statements are all true:
1. a higher dr cam has a con over a lower dr cam that it can resolve a higher latitude, but only if a certain scene exceeds the dr of the given low tier DR cam.
2. if a scene doesn’t exceed the dr of a given lower tier dr cam, and the operator has nailed his in-camera exposure, the higher dr cam wont resolve more detail
3. with a low dr cam you have less room for failure if you try try to map the dr of a given scene to the DR of the camera, compared to a cam with high dr
4. a good dop equals a good technician, and if you don’t understand the principles of DR you’d probably fail on about any camera

If you look at the Mavo LF footage of GAFPA GEAR, you will notice that we rarely clip highlights, because we hate clipping of highlights, we either avoid clipping by ETTL (exposure to the left), or avoiding certain daylight scenes. In our humble opinion to create a nice image in post your scene shouldn’t exceed 8-10 stops, and the rest is only used as roll off. If you fail doing so, you probably end up with a wrong looking image. Look at painters (rembrandt carvaggio), Rembrandt uses strong contrasts, by searching for deep non black and whites, and contrasting compositions, if you take your spot meter to the museum you will find out that the difference between the deepest and highest stop are not more than 6.5 stops (spotmetered this myself in the Rijksmuseum in Amsterdam), still Rembrandt depicted scenes that might have had about 10-12 stops in real life, but i guess not more, Our eyes tend to hate compositions that go over these values, and even though you can clamp these stops to a lower DR range in post , its a good rule of thumb to reject any scene that spot meters more than 12 in real life.
This is of course all subjective. The Edge gives me 1 stop more than the MAVO LF< which will free me a bit from having to choose the exact right exposure, and give me the ability to resolve some scenes that the MAVO LF wasn’t able to resolve, as I ve referred before , I was happy with the mid and highlights, but sometimes was opting for a bit more stretched blacks, something the edge will give me ( if I understand how to map the exposure). A big issue with high dr cameras is that its hard to understand to expose them the best, a low dr cam will be much easier to expose in that regard.

he LF seems to look more “finished but it could just be the LUT used. This isn’t a negative to me but a reality that more DR equates to a flatter starting point in most cameras which can mean more PP work in many cases if not using a reference LUT/preset.

I rarely (lets say never) use LUTS for LOOKs, only sometimes technical CST (color space transforms), when I first got the EDGE I had to spend some time to understand it, because every camera is different, and I also didn’t want to go to filmic in my approach, because having the EDGE as one of the first is also a bit of a responsibility. I don’t really believe in LOOKS, if your a good grader and a camera is good enough (it captures enough and good data, in a good codec), you can let it look like anything you want it too, as long as your a experienced grader. OFc some cameras might be easier to grade, because there are more luts available from third party suppliers, or the base grade (base matrix), is very close to how you want it to look like, and thus you only have to dial in some contrast. In my believe the EDGE can look exactly like the MAVO LF, (i will later on show these things with some side by side images), but the MAVO LF cannot lookexactly like the MAVO LF in a lot of situations, because the EDGE Captures more an better data. Over the coming months you will see a lot of EDge footage from our side. Watch out for new content on our Vimeo, Youtube or on our website.
https://gafpagear.com/kinefinity-mavo-edge-footage/

iiiNelson

Jerome

Yeah I’m a big proponent of ETTL myself on the photo side. Personally I kinda hate the washed out ETTR look that’s popular with a lot of people because that was the most consistent workflow on a Canon. On Leica or early Sony cameras it was always recommended that you need to protect your highlights and ETTL. Shadow detail can be brought up in post and the eyes/brains have a natural affinity to disregard deep shadows unless something stands out.

I’m already subscribed to your Vimeo/YouTube pages… and I’ve watched some of your content multiple times.

Jerome

iiiNelson
hey Nelson, I dont believe when executed int he right way that ettr will lead to washed out blacks, but i def see your point, in reality people dont know hwy they go ettl or ettr, they just something they heard on the streets and without further knowledge using it to create another garbage look 🙂 Ettr or ettl is often referred to as CINE EI, like in cameras like ARRI ALEXA, the analogue sensitivity isnt changed thus clipping point and dr is same, yet the digital curve of how the data is interpreted is changed so you see the data displayed differently and thats why you change the exposure either by nd, or shutter, or aperture. By doing so you shift the middle point gray over the dr range, shifting the amount of stops above middle gray and below. If you do this and grade accordingly in post nothing happens, except that you might gain a bit of highlight information (traded for less fidelity in the shadow department), or the other way around.

Jerome

SO in this fun ongoing topic, I want to shift the conversation a bit to artistry, because all these technical things to me only serve one purpose namely; making/creating interesting images and stories and worlds.
As I have mentioned many times before I believe in rec709 as a future proof format. even though the gamut might be enlarged to me rec2020 with its potential high dynamic range is too far off from my believes it will hurt the eyes and desaturate the heart. I was being raised with art, to start with paintings, from Caravaggio to Diebenkorn, to Rauschenberg to Malevich to Vermeer. One thing they all have in common even though some of these quick examples use mixed media is that their dynamic range of what they exhibit is never more than 6-8 stops (yes Rauschenberg had some installations with incandescent lights, but even these were kind of muted).

Diebenkorn

K Malevich

Bruegel

Rauschenberg

As we all know cinema is highly hijacked by capitalisms, and most innovations, are pushed by the big players (Sony etc), and not really the artists. The term ‘ immersive’ is often used by these players. in their opinion we need 3d, huge screens -where you cannot oversee the whole screen and instead have to move your head-, 5d (also wind ,water and odor). To them cinema should imitate life and not transpose life (transposing means we can see and feel and understand the world with different observations/angles)

Yes , what if these painters had richer blacks and whiter whites? Would they have used it? Yes probably, but not for good.
The issue with paint is that the only way to increase contrast is to decrease reflections in blacks, so that black appears deeper and thus white in relation more bright. Yet if we do a deep study to painters their needs were never to have more contrast but instead have more color variaty and textures, if we look at the development of paint, we see that the biggest innovations and talk abouts where the invention of more colors(pigments) and eventually the paint in tubes in 1841. The invention of paint in tubes could be compared with the invention of the portable camera. The idea that you could go into the field and bring your favorite colors and could just paint without having to bring an enormous amount of stuff to mix paint etc, was a huge revolution, it’s kind of what we saw with the canon 5d for aspiring filmmakers. All of a sudden the amount of poetic realism, documentary painters sky rocketed, leading to whole new styles of painting and eventually impressionism.
That said, innovations are always part of what drives artistry, but also limitations help artistry. Look at medieval paintings, they have in common that they only had access to a few different pigments to paint with (like red ochre), it’s very distinctive and some painters could really come up with superb compositions and color arrangements.
Raphael - the cardinal

So if we look at the history of painting, colors, textures were the main innovations and also the things the artists craved about the most.
If we look for instance at documentation about Rembrandt who made immersive paintings which seem to have very high contrast but just by depicting scenes with high contrast. Art is about seeming not being, by depicting the sun we can feel and see it’s strength just because we can imagine the sun.

Science wise a lot of research has been done after how we perceive colors and contrast, and it seems our brain doesnt like more than 8-9 stops of dynamic range. Once it exceeds our brain has to work hard, we have to capture the very bright parts of a scene by closing down, and capture the shadows by opening up to them and combine them in our brain into a reasonable image. But that takes effort.

To make a quick statement here are three images, sadly not very artistic but to proof a point. Spot metered we see that the sky is the brightest point at around 12.8 stops. and the darkest part in her hair at about stop 1. We can capture this full range with the mavo edge, ofc the shadows will suffer a bit from grainyness, and thus less resolution, but due to 8k capturing this is very fine and can easily be taken care of by advanced nr (which i didnt used in this quick example). If we start color grading and want the outside to be a normal exposure we will have to underexpose the inside. The other way around is ofc the same issue.
What we can do is start decreasing the contrast but since this math is applying to the full image it will start to look weak and the blacks will be lifted by a ugly amount washing out the image, same goes for the whites. A good image in general needs a white or black point to level up the rest of the image to.
In this example below we graded for the outside

in this example below we graded for the inside

Now even if we used the mythical roll off (currently the most used word in populist camera forums) , we still would not really benefit from our high dr camera. SO if you lack the ability to color grade (repaint), you wont need a high dr camera, because you are pushing this grade into a rec709 space, and to make it a bit good looking we have to either underexpose the foreground or overexpose the background. So even though our camera can capture it we will throw it away in post.
Filmmakers who have took lots of advantage over HIgh dynamic range with cameras like the alexa 65 are Emanuel lubezki and alejandro innaritu in the film The revenant, even though the film proofs to be mediocre from a artistic point of view (mine pov and maybe not yours), its still a tremendous achievement technically and from an image make point of view (if filmmaking was only about image making this film would be great) there are some truly beautiful frames. But what most viewers dont know is that these images arent shot in-camera.. they had to move fast, and couldnt use light sources, so at a very high level they did relighting in Autodesk Lustre, masking about every person out and changing the balance and colors. This post- production process took ages, but because of that they were able to use a more documentary kind of way of shooting which fitted that project. Ofc still a big crew, but due to the extreme weather and doing reshoots in other seasons light changed so much that there was no other way of doing so. So even though most images could have been shot maybe on a lower dr cam some scenes had simply too much dr, looking ugly to the eye, and had to be rebalanced. High dr cameras are great for if you are a painter and want to repaint your images in post. But there’s only one way and that is starting to use animated powermasks…
If your arent someone handy in the field of post production you probably dont need more than 10-11 stops of dr at max. As long as you set your camera at the right settings (lens, iso, shutterspeed, nd) it will do.
Grading the revenant

Now back to our simple thesis the shot we all know, person sitting inside and inside its much darker than outside, and we somehow want to retain the background (the world outside the window). By trying to add a light by means of a powermask with individual parameters we can start to make it happen, but our eyes are not so easy to be fooled sadly. Still this light coming from the outside is the keylight for the inside, so we have to understand when working in a grade we want to work in 3d space yet our images are in2d. This is where painting knowledge is needed, painting does the same, it seems to be a 3d world yet its established ona 2d flat canvas. So we have to study light and by doing so during a grade we will also start to learn about our own compositions, how we could have done better how we could have added fill light etc, and thats the reason I think every so called dop should learn the craft of color grading, to my believe you should be a bit reserved calling yourself a cinematographer if you just point and shoot, cinematography is image making, you dont need to necessarily manipulate your images in post, but you need to establish yourself as an artist and start discovering. I myself am not a cinematographer, I am only active as filmmaker and Ive learned from a few great dops ( for instance from the previous mentioned Emanuel lubezki during a shoot in 2009 which took two months and gave me the opportunity to learn a lot from his visions), but I believe in myself as a filmmaker/cineast in which i try to be both the cameraman and director and editor. So i know I will never be as good as those people or else I have to compromise on my other interests, so I have to be humble, and keep developing myself in a broad way. This forum helps me to write down these ideas and revisit them by sharing them with you.

I quickly made a few powermasks to balance the outside to the inside. On top I added (not included in the image down below) a powermask to highlight the person sitting on the couch. i did this with two power masks and follow the already existing keylight which lifted to a smaller degree than the fill light

Here is a small video showing you this issue and how we could possible resolve it.. shot with the MAVO Edge, and spot metered a 12.8 dynamic range.

These are just plain and boring examples, but sometimes these work the best. Our goal is to take you on our journey, and this topic will be crowded with a lot more interesting examples and actual projects we will be working on. Join the conversation!

Jimmy

Thank you for this very interesting conversation. There is the acronym “nr” which I have seen come back several times but which I cannot understand.

Jerome

Jimmy hey Jimmy! Welkome on our forum!
Yes! I should include a list of acronyms. I should revisit and also do something about my grammar. Generally I use the following acronyms:
NR : Noise Reduction
Ofc : Of course
DR : Dynamic range

Jerome

Revisited this topic, and have to conclude that Real life situations always tell me and most of the others the most, in that regard ‘ Dynamic range’ is more or less a conceptual technical term and ‘latitude’ an empirical more practical way to talk about this. I was just standing at station Sloterdijk in Amsterdam on the lowest deck, which is a dark alley /corridor. about 350m up in length, with a few open gaps from left and right, just enough for a tiny bit of fill.
The lights are old fluorescents balanced at around 3000 kelvin, and over day they almost have zero effect other then a tiny bit of orange glow and a bit of vibe. Sun was harsh and out in my direction, and I was waiting for my train, and I try to study light whenever I can see, why it has a certain quality, and I always try to bring a camera, even though it might give me a hernia some day. Even though I knew the camera would succeed I wanted to see what the camera would do with this light situation. basically not a lot of mids, but a bright outside with the sky and the white stones, and inside a bounced grey environment with a few incandescent bulbs. My eyes tend to have to focus either to the outside to avoid clipping that or, open up for the inside which is dark and non contrasty. If a camera can capture this, and still have enough latitude to open up for different grades, I would be a happy man, because I tend not to be willing to capture stuff that’s unpleasent to my eyes, Because everything starts with the eye! Any way I should stop talking, Here are a few images and I included a dropbox link to a somewhat uncompressed DRX, please note that this is not a detail test, this was shot with a tiny mamiya sekor 58mm 135 lens, and an aivascope 1.5X anamorphic adapter.
graded

graded + analogue grain and halation

log, no effect

Dropbox for dpx log file:
https://www.dropbox.com/s/hn9x7rl8j1lx085/anamorphic%20edge%20log_1.44.3.dpx?dl=0

Eric

Hi Jerome, I just got my Mavo Edge yesterday and trying to figure out some of the features in the camera. There’s a ‘Tune Highlight’ feature that new to me, I tried to crank up the numbers.. It helped on highlight details but get noisy in the blacks. What’s the best setting for this number when I’m shooting an indoor scene with bright outdoor window in order to get best DR? or is this number determine by highkey or lowkey lighting in the frame? One more question regarding the dual ISO setting, is this a auto switch kind of setting when we crank up the ISO, it will push to the 2nd high ISO setting? I couldnt find any ISO setup for dual setting. Your advise would be much appreciated. Thanks.

Jerome

Eric Hey Eric i merged this question in to a wider conversation since it matches this topic (dynamic range).
We advice you to start with sensitivity mode to ' Cine EI" . Basically this gives you the best Dr and a very easy way of mapping your dynamic range in a convenient way, in such a manner that you can easily monitor your highlights and shadows.
Shooting in a very wide contrast scene and you want to favor the highlights? Set your zebra to 99% unclip the data you want to retain, and then crank up the cine ei level, and see if your shadows hold up niceley, if not then compromise by slightlyu sacrificing and clip some data. Basically cine ei let you set the middle gray ppoint without changing clipping points. But be warned once you get to 1600 iso in cine ei you automatically switch to the 2nd base iso (2560), in sensitivity mode iso, this is the same point, and with iso mode the behaviour of the camera is more complex and in my opinion not better. So try cine ei first and let us know if you understand the concwept, if not, there are many documents on the web showing you more info about this for instance this one:
https://www.abelcine.com/articles/blog-and-knowledge/tutorials-and-guides/iso-ei-and-gain-explained

Eric

Jerome Hi Jerome, I tried the ei setting as your suggestion.. use zebra for highlight protection and crank up the ei to get more details in shadows. But the mid and dark shadow areas seems to have quite prominent noise even at just ei640-800. I haven’t shot anything to download and check on bigger monitor, but judging from my 7” SmallHD onboard monitor.. it’s already quite heavy noise. 8k FF setting have more noise. Another issue I encountered is the dark streak line smearing out from bright spot in the frame, something like a reverse anamorphic flares which I think it’s hard to be fixed in post. Is this something I can solve with black balancing or this is the weakness of this sensor? Please refer to the photos I have attached for reference. I owned Varicam LT and S1H as well, shot quite frequently with the high iso setting, the footages always seems usable and reasonably clean at iso 4000-5000. But I’m quite disappointed seeing the noise in Mavo Edge at the moment, not sure is this something that can be solve by black balancing or firmware update. Looking forward for your answers, and thanks in advance for your help

[upl-image-preview url=//i.imgur.com/19CWuCg.jpeg

]