5D Mark III raw video (Magic Lantern) versus KineRAW MINI for image quality

[Andrew Reid]

The €3199 KineRAW MINI is the latest affordable raw cinema camera to be released. Featuring a 4K Super 35mm sensor it is a rival to the Blackmagic Production Camera and an alternative to shooting raw on the 5D Mark III with Magic Lantern. It shoots 2K / 1080p Cinema DNG uncompressed raw internally.

Having now spent more time with the camera for the full review (coming soon), I was curious to see how 5D Mark III raw video stacks up against a dedicated cinema camera with a sensor purpose built for video - not stills.

With the help of a new EOSHD test scene, we're about to find out -

[url=http://www.eoshd.com/content/11462/5d-mark-iii-raw-video-magic-lantern-versus-kineraw-mini-image-quality]Read the full article here[/url]

[andy lee]

it looks like it is a very good little camera producing a very good image - one of the best so far....

[QuickHitRecord]

I'm just floored that Canon have made one of the most accessible cinema cameras out there -- and they weren't even trying.

[versuviusx]

still the canon in raw mode does not do recording.

this is a big deal for me.

all the raw image videos on the net do not have sound.

they take the video and apply music and thats all you get.

cant wait to see how the bmcc 4k competes against the 5dMkIII raw.

but still no audio for the canon. its a deal breaker.

[Germy1979]

Awesome review Andrew! I was wondering at the end how long you might have the Kine, lol. It'd be interesting just to see a range test
between the 2.5k BMCC and the Kine since they both rate around 13 stops? Noise, highlights, whatever. I'm sure writing a whole review is out
but that's about the only thing I would be interested in knowing the difference. Especially lowlight where the bmcc just pales in comparison to the
Mark 3 raw.

[Shaun Fontaine]

still when it comes down to brass tacks, the KineRAW MINI does what it is suppose to do without a hack. It was built for purpose. The 5DM3 is amazing for what it does, it always seems no matter what test is done, the 5DM3 (RAW HACK) always seems to come out ontop, even against dedicated RAW video cameras such as the KineRaw and those by blackmagic. But when it comes down to it, it's not canon is it, it's Magic Lantern who is to thank. Now i'm not anti hack but one thing is for sure, Canon will never give us real pro video features at this price, they will never give you RAW or Prores or DNxHD or anything like that, so hacks will always be the way people have to go. In the end, better and cheaper cameras by the likes of BMCC will come out and eventually that line of 5D's will become obsolete for those who shoot for film making purposes who want quality codecs and RAW and 4k and so on. I mean even the Blackmagic 4K camera is nearly there at that point. Ok it only has a s35 sensor but who cares when it has global shutter and 4k images. All I'm saying is, Canon need to change or eventually people won't be buying the 5D series anymore for Video purposes in the future years to come.

[yohh770]

would love to see more ......

[yohh770]

basically its about color and sensor size. Only 3 companies can do it  .....sony, fujifilm, canon...

[zephyrnoid]

basically its about color and sensor size. Only 3 companies can do it  .....sony, fujifilm, canon...

....but Panasonic has something new up it's 'sleeves' so we shall see.

[zephyrnoid]

Great Article and comparo- though a 15 step gray strip would not kill you Andrew ;)
"The difference in look between full frame and Super 35mm you know already. All else equal you get a shallower depth of field on the 5D Mark III and a heavier fall off in brightness towards the edge of the image with lenses when shooting wide open. This can be a very attractive look. It’s also easier to isolate the subject from the background when using a wide angle lens on the 5D Mark III (i.e. 28mm) than it is on the KineRAW MINI (18mm)."

FF sensor may indeed yeild "an attractive" look IF only you can hold the main subject in critical focus in shots with moving actors, locomotion etc, which is the principle reason Super35/DX/APS-C are the ideal compromise for focus leeway with M43 absolutely too difficult to get anything out of focus unless you are using very long focal length lenses.
So frankly, I keep chanting the Super35/DX/APS-C mantra :wub:

[ScreensPro]

Interesting, did you test any of the 720p slo-mo modes?

[Guest]

If anyone can be bothered, I'd really appreciate having the relationship between dynamic range and ISO performance explained to me (I do understand the two concepts independently btw). 

 

For example: If Andrew's test scene above was considerably less well-lit, yet still designed to benefit from 13 stops DR to show all main detail, which camera would "win"?

 

I know the answer is subjective to some extent, but what I can't fix in my head is how the 5D can be the better low-light performer and the Kine have the greater DR. Surely a better high-ISO image also means better in the shadows, and therefore can be lifted more if you expose to the left (as in first part of Andrew's test where he keeps the highlights on the Super-8 cam with the 5D). In that part of the test the Kine wins, and yet in the low-light stakes the 5D wins! Arrrggh I don't get it! Is it just that the 5D is better all-round (colour etc) in low light? 

 

I hope this question is clear enough for you to understand what I'm asking - I am confused so it's not easy to put into plain english ...

 

[ScreensPro]

Matt, it gets even more confusing, i'm afraid... as most camera generally achieve less DR as you go away from the "native" ISO of the sensor... So shooting high (and low) ISOs usually wont get the max. DR from the sensor (though some cameras seem to maintain DR across the ISOs, the Canon C range, for example)

 

re: ISO and raising the shadows, it does help, but only goes so far. The sensor can only record the detail within the DR it is capable of... So no amount of raising the shadows or dialling back the highlights will get back detail that wasn't there to begin with.

[tobyloc]

If anyone can be bothered, I'd really appreciate having the relationship between dynamic range and ISO performance explained to me (I do understand the two concepts independently btw). 

 

For example: If Andrew's test scene above was considerably less well-lit, yet still designed to benefit from 13 stops DR to show all main detail, which camera would "win"?

 

I know the answer is subjective to some extent, but what I can't fix in my head is how the 5D can be the better low-light performer and the Kine have the greater DR. Surely a better high-ISO image also means better in the shadows, and therefore can be lifted more if you expose to the left (as in first part of Andrew's test where he keeps the highlights on the Super-8 cam with the 5D). In that part of the test the Kine wins, and yet in the low-light stakes the 5D wins! Arrrggh I don't get it! Is it just that the 5D is better all-round (colour etc) in low light? 

 

I hope this question is clear enough for you to understand what I'm asking - I am confused so it's not easy to put into plain english ...

 

 

I think, looking at Andrew's comparison, the Kine has more DR, probably 1.5 stops more at (and this is key) low ISOs, i.e. ISO400 where they both perform with little noise in the shadows. However the 5D is about a stop cleaner as you bump up the ISO so my guess would be if you ran Andrew's test shot where you can see the limits of both cameras, the highlights and shadows at the same time but this time shot them both at ISO1600 that Kine advantage in DR might disappear because the shadows would be noisier.

 

This I think is partly the advantage of a full frame sensor, it tends to be a stop better than a S35 sensor in low light as it collects a stop more light over it's surface. That's why I personally like FF sensors more than S35, because you can chose to have the exact same DoF as a S35 camera by stopping down one more stop on the lens at no noise penalty BUT you also have an option for shallower DoF if you'd like AND crucially, lenses that are a stop slower are cheaper, sharper, easier to make and if you're lucky come with IS.

[markm]

Thanks Andrew!

[Christina Ava]

also the KineRAW MINI looks like a spare part from the Kaypro II,

i wonder why they dont invest in the design at all.

who would buy something that looks like it came from a time capsule of the 50s.

 

[Guest]

I think, looking at Andrew's comparison, the Kine has more DR, probably 1.5 stops more at (and this is key) low ISOs, i.e. ISO400 where they both perform with little noise in the shadows. However the 5D is about a stop cleaner as you bump up the ISO so my guess would be if you ran Andrew's test shot where you can see the limits of both cameras, the highlights and shadows at the same time but this time shot them both at ISO1600 that Kine advantage in DR might disappear because the shadows would be noisier.

 

This I think is partly the advantage of a full frame sensor, it tends to be a stop better than a S35 sensor in low light as it collects a stop more light over it's surface. That's why I personally like FF sensors more than S35, because you can chose to have the exact same DoF as a S35 camera by stopping down one more stop on the lens at no noise penalty BUT you also have an option for shallower DoF if you'd like AND crucially, lenses that are a stop slower are cheaper, sharper, easier to make and if you're lucky come with IS.

 

The sensor can only record the detail within the DR it is capable of... So no amount of raising the shadows or dialling back the highlights will get back detail that wasn't there to begin with.

Thanks guys, that's helped. SP's point that a sensor's DR is fixed whatever you do to the image is probably the crux of it. I guess my question was similar in that way to the RAW vs flat-profile-compressed DR debate. 

[zephyrnoid]

LOL!
"who would buy something that looks like it came from a time capsule of the 50s."
I would! Today's my birthday and I love the 1950's . Your mistake is that Kaypro was a way past the 50's. But I could care less what it looked like- it simply has to perform. As I teach in my classes
Looks BEHIND the lens don't count. Looks only matter in FRONT of the lens

also the KineRAW MINI looks like a spare part from the Kaypro II,

i wonder why they dont invest in the design at all.

who would buy something that looks like it came from a time capsule of the 50s.

 

[zephyrnoid]

If anyone can be bothered, I'd really appreciate having the relationship between dynamic range and ISO performance explained to me (I do understand the two concepts independently btw). 

 

For example: If Andrew's test scene above was considerably less well-lit, yet still designed to benefit from 13 stops DR to show all main detail, which camera would "win"?

 

I know the answer is subjective to some extent, but what I can't fix in my head is how the 5D can be the better low-light performer and the Kine have the greater DR. Surely a better high-ISO image also means better in the shadows, and therefore can be lifted more if you expose to the left (as in first part of Andrew's test where he keeps the highlights on the Super-8 cam with the 5D). In that part of the test the Kine wins, and yet in the low-light stakes the 5D wins! Arrrggh I don't get it! Is it just that the 5D is better all-round (colour etc) in low light? 

 

I hope this question is clear enough for you to understand what I'm asking - I am confused so it's not easy to put into plain english ...

This is why I chided Andrew for omitting a 15 step gray strip. It would answer your question graphically.
DR is about the number of VISIBLE and measureable discrete 'values' that can be recorded and displayed ( or printed ) , inclusive, from 100% Black to 100% white so a 13 step range is 100%Black, 100% white and 11 other values in between.
Shop around Google search for more in this.
Low light sensitivity is completely separate.

[Guest]

This is why I chided Andrew for omitting a 15 step gray strip. It would answer your question graphically.
DR is about the number of VISIBLE and measureable discrete 'values' that can be recorded and displayed ( or printed ) , inclusive, from 100% Black to 100% white so a 13 step range is 100%Black, 100% white and 11 other values in between.
Shop around Google search for more in this.
Low light sensitivity is completely separate.

Yes that makes sense, thanks. I like that Andrew is using a "real world" test setup but a 15 step gray card would certainly have avoided my confusion. Sometimes pure technical information is important ...