Raw samples of Canon XF300 & Sony EX1R

[Les Wilson]

Originally Posted by Brian Woods

I have to respectively disagree with Alister's comments on the ergonomics of the EX1 winning out over the XF300/305. I can shoot all day (and did yesterday) hand held with the xf300 and experience only a fraction of the wrist and arm fatigue I get after spending an hour with the EX1. The balance is much better on the XF300. The EX1 is a near nightmare to use hand held for significant amounts of time due to how far the handgrip is from the camera's gravity center.

However, like he mentioned, it will take some getting used to throwing a switch every time I want to go from using the zoom rocker to using the zoom ring on the lens.

Ergonomics of cameras is far more than this. In the case of the zoom operation, I don't think it's preference. The cognitive load to operate the XF zoom strikes me as arguably higher to switch modes as well as maintain the state of rocker vs lens ring since the XF design makes them mutually exclusive. Cognitive load spent on operations is cognitive load taken away from the creative.

Another example is the EX auto iris on the grip. It offers similar low cognitive load benefits as it lowers the number of things a busy left hand must perform and shifts it to an otherwise unused finger on the right. I think it's arguably also a more reliable move (less load) with the right hand pinky than groping with the thumb or *gasp* complete cessation of zoom and focus operations to use the pointer finger.

On the grip, I wish someone would do a good evaluation of the balance/fatigue issue including using the improved grip on the updated EX1R model released last fall where they shifted the location of the palm and location of the strap. It's not a marathon of which can be held the longest. There's also the functional issue of the rotating grip of the EX that allows positions and moves not possible with the XF design...so it's a tradeoff.

Alister makes some other astute observations about smaller things such as the XF design requiring the operator to remember where manual focus was last set and the lack of WFM in the VF. There are also things in the EX that don't exist in the XF and are missed in reviews looking only at the XF (eg, tripod mount, up to 15 seconds cache record, rotating handgrip, WFM in the VF, labeled f-stop markings, hard stop iris ...).

Size matters. The XF is huge in comparison to the EX. The body is nearly 3 inches longer (not counting the VF), 1.4 inches higher and nearly a pound heavier. This plays in ergonomics well beyond the hand grip.

On the image side, I read the numbers thrown around and debated but I see this example of the EX1 at f3.4 and the FX at 2.0.
http://www.repaire.tv/tests/alicefor...el%20diaph.png

and this with both at +12db:
http://www.repaire.tv/tests/alicefor...20open+AGC.png

and I read comments about tweaking the XF NR or whatever to compensate and I think only how much harder the XF is to operate...ergo worse ergonomics

[Peter Moretti]

I have no doubt that the EX1 is better in low light, but the WB for the XF was clearly off in those images. The 5600K preset did not make white white, so a custom WB or other preset should have been used. And this would make the XF image somewhat less muddy looking.

[Brian Woods]

The first test grab makes sense, and completely corroborates my notes. Shooting in progressive modes we found f/4 on the EX1R to match well with f/2.8 on the XF300. The Repaire.tv tests used interlaced modes for their tests, and the EX1R does get a nearly half stop sensitivity boost when shooting interlaced. The Canons, sadly, don't see such a boost - however, I never have the need to shoot interlaced anymore, anyway, so it isn't a big issue.

The 2nd grab - They are using AGC in that comparison, and it looks like they labeled the Canon at 16.5 db? Not really sure what's going on there. The NR does work quite well on the Canon, and setting it to "Auto" takes the guess work out and gives quite pleasing results at higher gain values (about 2.5 db noise improvement).

Cognitive load is... whatever it is. Once you learn a camera, you can operate it like the back of your hand, and they're all different. I've never picked up a new camera that I didn't have to take some time to learn, and I've never met a camera that was so complex I couldn't chew gum and walk at the same time with it. And I've never used a camera that didn't require both hands... As long as I don't have to go through a menu, I'm happy ;) What I like about the Canon is that it does have 13 user-assignable buttons (2 of them on the right index finger above the record button), so if you don't like where something is, put it where you want it. I've never had a problem with the Sony button layout, seemed pretty intuitive. Obviously easy access to auto-iris is important to you; for me, I very rarely use it.

As for ergonomics, I guess it all comes down to personal preference anyway. Usually when I'm using a camera like this, it's for a long time and it's all hand held (reality shows & documentaries) - so yeah, for my usage, it's a marathon of which can be held the longest. The XF camera is heavier and longer, but it's so much better balanced than the EX1R that I'm able to comfortably operate it for much longer than I can the Sony.

The XF does have cache recording, although I've not used it yet and can't say how long it's for. The rotating handgrip is for me the cause of my major ergonomic issue with the Sony, so I don't miss it one bit in the Canon. A hard stop iris would be nice, though.

Anyway, the Sony does edge the Canon out in low-light performance, but not by nearly as wide a margin as one would expect. Like any camera, it takes learning and getting to know it, and once you do it competes very favorably against the Sony EX1R (and the vice versa is also true - it ultimately comes down to personal preference & individual need). I would see no challenge for anyone using this camera for weddings, events or news - after all, that's pretty much what it's been designed for.

[Brian Woods]

Originally Posted by Peter Moretti

I have no doubt that the EX1 is better in low light, but the WB for the XF was clearly off in those images. The 5600K preset did not make white white, so a custom WB or other preset should have been used. And this would make the XF image somewhat less muddy looking.

I believe the Repaire.tv tests were done with a pre-production camera. Many features and customizations were not enabled on the pre-pro models, and their picture had a slight maroonish tint to them.

[Alister Chapman]

My Ergonomic judgements were made between the XF305 and EX1R. The EX1R is quite a lot nicer to hold than the original EX1.

[Tans Mark]

Dear Alister,

Thank you for the XF305 vs EX1. To be honest this test is confusing for me. The Xf300/305 and EX1 samples I had seen so far (from various sources linked to various thread in this forum) showed that the XF300/305 has sharper and more detailed image.

Your test showed that in most cases the EX1 has sharper / detailed image.

So who is right at the moment ? Until know it was clear that the XF is better, but you showed that EX1 is better.

[Alister Chapman]

Don't confuse the very fine noise from the XF305 with true picture detail. It took me a while to work out what I was actually seeing, but the extremely fine noise and grain that exists throughout the XF305's footage can give the impression that there is more detail in the 305 footage than there really is, especially when viewed as a frame grab where the noise is static. The grain tends to add a fine texture to everything. In my tests I was looking at the foliage and trees and thinking the XF305 was resolving more detail as the foliage appeared to have more "texture". But closer examination shows that there is no more real detail than from an EX1, what I was interpreting as detail and foliage texture was in fact noise.
I would expect both cameras to have almost exactly the same resolution as both are using 1920x1080 sensors with a good Bi-refringent filter to control aliasing, thus both cameras should resolve almost exactly the same. Looking at the same clips I then looked at the car and brickwork. The car is very revealing as the 305 noise almost adds a texture to the smooth panels which remain clean and smooth on the EX1R. At the same time the registration plate on the car is easier to read in the EX1 shot than the XF305 and the brickwork shows more true subtle detail in the EX footage, some of which is being masked by noise in the 305.

The differences are largely down to noise, noise reduction and differences between the detail correction and sharpening circuits used in the cameras, but the very fine XF305 noise can be somewhat misleading making some parts of the image look more detailed than they really are. You can simulate the effect in Photoshop by taking an image and using the "add noise" filter to add fine noise to an image.

[Laurence Janus]

Many thanks for the review Mr Chapman, the most informative yet.
I hope you can get your hands on one for longer next time.

[Tom Roper]

I've used Imatest to measure the resolution of the EX1. It can resolve all the way to the nyquist limit, more than 1000 lines. I was testing at the center, there could be differences at the borders and at different iris openings, but I would expect that any major difference in the sharpness comes down to processing of the detail and noise circuits. If one cam should appear markedly sharper than the other, it probably owes to processing. You should be able to achieve the desired sharpness look without aliasing, up to the limit of 1080i/p format.

[Brian Woods]

Originally Posted by Alister Chapman

The car is very revealing as the 305 noise almost adds a texture to the smooth panels which remain clean and smooth on the EX1R. At the same time the registration plate on the car is easier to read in the EX1 shot than the XF305 and the brickwork shows more true subtle detail in the EX footage, some of which is being masked by noise in the 305.

The differences are largely down to noise, noise reduction and differences between the detail correction and sharpening circuits used in the cameras, but the very fine XF305 noise can be somewhat misleading making some parts of the image look more detailed than they really are. You can simulate the effect in Photoshop by taking an image and using the "add noise" filter to add fine noise to an image.

Thanks for the side by side tests, Alister. My observations looking at the CP & PP versions:

The car seems to show much more aliasing than in the PP EX1 shot compared to the 305, especially when viewed at 200%. It looks like the PP you use really increases sharpness and adds more aliasing and no more actual detail. The brickwork on the right side of the framegrab, maybe due to the noise in the 305, looks much more even & detailed in the 305 shot. I don't see more true subtle detail in the EX1 shot - in fact, it's quite the opposite. I see a lot of areas where the EX1 smoothed over fine detail and created big blocks of solid brown, whereas on the 305, when viewed at 200%, it looks like I can see the white cement between the bricks without any of the "smoothing" or "blockiness" that the EX1 frame has.

In the default versions, everything is pretty even except a l noise in the sky in the 305 shots, but also more shadow detail (again, in the brickwork & buildings to the right).

The noise of the default grabs is an interesting issue because clearly there is a more of it in the 305 shot, and in framegrabs it doesn't seem ideal. However, in motion this kind of noise would never be a problem - there really isn't that much of it anyway. Taking the sharpness level to -3 seems to mitigate most of. At least here in LA, the first thing every post house does when they get digital footage is add noise to it, anyway - especially Sony XDCAM footage, which can have a "metallic" smooth look to it.

[Brian Woods]

And before anyone jumps on me about this, here is what I'm looking at (taken from the CP/PP grabs). There is definitely more fine detail in the 305 shot than in the EX1 shot, and it isn't an illusion of noise. Look at the brickwork and you'll see smoothing and blockiness in the EX1 shot. Also pay attention to the white siding at the top compared to the 305.

Just a disclaimer - this is viewing at 200%, and is heavy nit-picking. But if you want to find a difference between the cameras, this is where to look.

[Alister Chapman]

Originally Posted by Brian Woods

At least here in LA, the first thing every post house does when they get digital footage is add noise to it, anyway - especially Sony XDCAM footage, which can have a "metallic" smooth look to it.

Really? Why? Noise makes footage harder to grade, eats codec bandwidth, messes up narrow bandwidth television broadcasts and interacts with TV sharpness settings to potentially create a real mess. The very last thing you want to do is add noise. That's a terrible way to treat any material.

[Les Wilson]

Originally Posted by Brian Woods

...Cognitive load is... whatever it is. Once you learn a camera, you can operate it like the back of your hand, and they're all different. I've never picked up a new camera that I didn't have to take some time to learn, and I've never met a camera that was so complex I couldn't chew gum and walk at the same time with it. ...What I like about the Canon is that it does have 13 user-assignable buttons (2 of them on the right index finger above the record button), so if you don't like where something is, put it where you want it. ... Obviously easy access to auto-iris is important to you; for me, I very rarely use it.

As for ergonomics, I guess it all comes down to personal preference anyway.

The XF does have cache recording, although I've not used it yet and can't say how long it's for.

A hard stop iris would be nice, though.

Like any camera, it takes learning and getting to know it, and once you do it competes very favorably against the Sony EX1R (and the vice versa is also true - it ultimately comes down to personal preference & individual need). I would see no challenge for anyone using this camera for weddings, events or news - after all, that's pretty much what it's been designed for.

Simply defined cognitive load is the amount of mental "energy" something takes. Talking on the cell phone while driving is a demonstration of the impact to driving skills caused by the cognitive load of talking on the cell phone... driving skills drop in various amounts for different people and individual circumstances due to cognitive load.

Ergonomics does not come down to personal preferences. There are good and bad ergonomics. Having 13 assignable buttons versus the EX1R's 4 isn't necessarily better ergonomics, it's a functional characteristic (feature). Like function keys on a computer keyboard. They are ergonomically neutral. Like you point out, they CAN be used to improve operational ergonomics. And, the buttons themselves have ergonomic issues such as easiness to locate, tactile feedback, engraving, grouping etc. The membrane buttons on the original EX1 under the handle had poor ergonomics because of the poor tactile feedback of membrane buttons and they were changed to real buttons on the EX1R. Canon has legendary build quality and didn't make that mistake...kudos to Canon.

You can test a camera's operational ergonomics against various populations and evaluate the results to determine which designs are more reliable, faster, easier etc for various operations...whatever is important. Alister looked beyond the imaging block to other characteristics and took a decently broad look at various ergonomic issues beyond marathon holding and zoom mode switching....kudos to Alister.

The auto iris button is a great example of ergonomics. Sony located it in the same place it exists in most professional lenses. This facilitates ease of learning for those used to it there. Having available assignable buttons in another location is flexible but not necessarily better ergonomics. Right away, you have confusability issues finding the correct one (more cognitive load), reliability issues (how many times to people make mistakes, familiarity issues (relearning it after not using the camera for a time) efficiency ... all these things can be tested...again, it's not a question of functional equivalence, some approaches are ergonomically better than others....cognitive load often is a good indicator but not the only one.

The iris ring is another example. They both have them but the EX has stops and Alister felt the XF's was sluggish. These characteristics affect how hard/difficult they are to operate to get the desired results. The harder they are, the more cognitive load they take to operate and the less cognitive abilities you have available for other things (like driving haha). Taking your hand off the zoom ring on the XF to switch modes is another issue. The XF has smooth slow zooms (reliable, easy whatever) the EX1R doesn't (twitchy, stutters at slow speeds etc).

The XF cache record is for 3 seconds. The EX1R's is programable up to 15 seconds. These are functional characteristics. How they are activated is ergonomics.

Ergonomics was a perfectly valid thing for Alister to evaluate and make a conclusion about. It may have no value to you but it's not personal preference.

But I think we're off topic of this thread at this point. There's a thread for likes and dislikes that might be better:
http://www.dvinfo.net/forum/canon-xf...w-have-xf.html

[Brian Woods]

Originally Posted by Alister Chapman

Really? Why? Noise makes footage harder to grade, eats codec bandwidth, messes up narrow bandwidth television broadcasts and interacts with TV sharpness settings to potentially create a real mess. The very last thing you want to do is add noise. That's a terrible way to treat any material.

It's mostly houses that work on narrative shows with digital aquisition, where they have the budget for that sort of thing. They do it with RED footage, DSLR footage - basically anything that's not 35mm that they want to look more like 35mm

[Alister Chapman]

Adding noise is strictly taboo in the UK as it messes up the narrow bandwidth transmissions used for Digital TV broadcasts. The broadcasters go to great lengths to filter out any noise prior to compression for broadcast and as a result any noise tends to degrade the overall broadcast image.

Brian, going back to the brickwork, I concur with your observation that the XF305 handles the brickwork better than the EX1R. There is some blocking in the EX footage that is not apparent in the 305 footage around the darker coloured bricks. A lot of the "texture" on the bricks with the 305 footage is noise and varies greatly from frame to frame there is some blocking on the lighter coloured bricks, but not the darker ones with the 305. I'm not sure whether what your seeing on the car is aliasing or a result of the 4:2:0 color sub sampling, possibly it's a bit of both. It's quite interesting to see the subtle differences in the codecs behaving differently with different coloured bricks. It's not really all that surprising that the 35Mb/s codec of the EX1 is not quite doing such a good job as the 50Mb/s of the 305.

The EX1R with NanoFlash at 50Mb/s does not produce any blocking on the brickwork and is also less noisy, at 100Mb/s is really nice, but that's not what we are discussing here.