5D3 Hands-on Impressions and Quick Review

This quick review is purely textural. So, please don't ask me for any picture or further proof on what I say, as I will not provide these. Do bear in mind that things are being subjective as they are my own impressions mostly based on real-world experiences and no extensive measurement was carried.

Pre-requisite: The operation of the 5D3 maybe a bit complicated and it maybe especially difficult for new users. I assume that you should have some basic knowledge about its operations for what I am talking about in the following. If you have not used the 5D2/3 before, you may wish to view the 5D3 operation/user manual beforehand, which can be got here (click to download).

Build Quality: Frankly, it feels somehow cheap in my hands for the coarse pattern matte finishing of the outer shell of the whole body. When it is compared to the original 5D1, the difference is huge and the oldest 5D does feel more refined for the build. Unlike the 5D2, all the buttons are also now also matted, too, except for the shutter release button, which is yet still easy to get scratched anyway. The most unfortunate thing is the new buttons are rather cheaply built (they are smaller and the reaction force is terribly weak!) and thus the overall build quality feel of the whole camera body is also affected! Furthermore, the marks painted on the buttons seem to be easily peeled off, for what I could see there has already been sign of, which I have never seen the same thing happened with any of my Pentax DSLRs, in contrast! :-o All in all, the 5D2 and the original 5D are surely two better built bodies than the new III body, undoubtedly! >:-(

Shutter Action: Each time the 5D3 fires, it reminds me of an old traditional mechanical-type typewriter of the 80s for the sound and feel! Believe me, its stroke is the same! Really terrible! >:-[

Shutter Lag: Fortunately, the shutter lag seems to be better, although it is not considered as something lightning fast, unlike that of the EOS 1 series or those Nikon high-end bodies.

User-friendliness: The operation is 5D-ish, nothing special indeed. However, in particular, I don't like the idea of directly input exposure compensation using the rear quick dial, which was the design since the 5D2. The problem is that it can be accidentally turned whenever the shutter release button is half-pressed. Yes, the dial can be locked, but normally I should not lock it because it need to be used frequently. Still, I prefer the 5D1 design, i.e., to press a button and then turn.

The New "Intelligent Viewfinder": The focusing screen is an traditional optical type with grid lines and central spot metering bracket carved, overlaid with another layer of transparent monochrome LCD for the AF marks and the exclamation mark warning signal display. The craved and the LCD displayed marks are then illuminated with a weak red light source, altogether and for the whole screen. This illumination method looks rather cheap for outcome and the illumination is obviously not even. Practically, I think it has little meaning neither, except in a very dark environment where the user can see almost nothing anyway! So, after all, I just opted to turn the red light off!

Btw, the default 61-point AF indication is too crowded and clumsy and I have turned them all off until they are chosen and they should appear all the time if manually chosen. For auto selection of the wide 61-point, I set the camera in a way that the selected points will only appear after the focus points are selected and the AF is achieved.

Last but not least, it seems that the focusing screen of the 5D3 is not replaceable and for those who wants the previous S-series precision screens (EE-S for 5D1 and EE-G for 5D2) for higher manual focusing accuracy with fast primes, it will yet be another major disappointment.

Metering accuracy (Evaluative): Metering results are bright but not conservative enough, resulting in highlight clipping for many areas in frame in many situations. I understand why Canon make the evaluative meter to expose brighter and has a tendency of this, as the sensor seems not to be too good in retaining shadow details (seems to be worse than the K-5, honestly)! I shall talk a bit more about this in the IQ part below.

By default, the "auto lighting optimizer" is turned on with moderate strength at two steps, which can be turned off for more faithful brightness of the pictures to reflect the true scenes, providing that the metering and exposure is accurate by itself.

Metering accuracy (Centre-weighted Average): Metering results are a bit less bright in general but it has no intelligence of any in detecting any unusual situations, say, high/low reflectance of objects and backlit. In general, it is more consistent in behaviour and is more predictable and thus maybe more reliable. I believe the CWA metering of the 5D3 is yet brighter than those of most of the old traditional film cameras, nevertheless.

IQ (Colour Accuracy/Tendency): Its colour reproduction and rendition (with all L lenses used) is surely Canonish, with saturated colours but less tone details (micro-contrast) reproduced, which I do not like actually. As usual, the Standard mode is more contrasty, produce "darker"/deeper colours and tones. Say for example, this is easily noticed with the green colours, which is more like dark greens than the natural greens with some yellow colours and lighter greens in them. To get "better" colours, I prefer the Faithful or Neutral Picture Styles. The Faithful mode produces an overall a bit warmer tone whilst the Neutral which I believe is "whiter", but yet the colour renditions of both are found to somehow more natural than Standard.

IQ (Textural Feel and Per Pixel Sharpness): Skin texture rendition is surely plasticky, which is one of the reasons I don't like to use Canon gear to do portraiture. I have the feeling that 5D3 pics are being even more plasticky than the 5D2, maybe there is always more noise reduction applied (I set Auto). Per Pixel Sharpness is not that superb as expected for a brand new model without any pixel-count increment. And even using the most expensive Canon L primes won't help, e.g., the EF85/1.2L! :-( After all, if more sharpness (sharpening) is applied, obvious digital artifacts can be seen at pixel level. Shooting RAW maybe better, I don't know, but I hate bad JPEG engine in-camera by all means!

Auto White Balance: I am really disappointed with the AWB of the 5D3. The indoor performance under tungsten is not any better than my old original 5D1. The outdoor performance is better but it is not considered as being very stable (which can be noted for the changed skin tones across different frames taken under similar situation at nearby time). The tungsten/artificial/mixed light AWB of the K-5 clearly wins over the 5D3 there, really. It seems that these years Canon has not done much to improve their algorithms, they used to lead others miles away in terms of AWB reliability of their DSLRs years ago but now it seems that they are lagging far behind and being worse than Pentax! >:-( (Frankly, Pentax is the best system for AWB which is even better than many ILDCs!)

IQ (Noise): The advantage of 5D3 can be seen at above ISO 1600 than the K-5 and 5D2, with more details/better resolution rendered and less clipping in the red colour channel. Below that, the difference is not huge, frankly. But there is room for improvement for the K-5's metering system, which is not that accurate in many cases.

IQ (DR): The 5D3 is weak in highlight reservation but yet the shadow details are not retained/reproduced very well. In order to preserve both, the contrast is needed to be lowered, but then the images will look dull as a result and IQ suffers. The DR of 5D3 is not much different from that of the 5D2, which is yet weaker than the K-5 for what I can see. So, what's point of carrying a FF if the DR could not be better? (Besides the option of less DoF, higher optical resolution and a larger viewfinder view, which especially useful for ultra-wide angle shooting.)

AF Operation (User-friendliness): The AF zone/point selection options maybe too many, but yet these are not really practical than I wanted. I recommend to enable only those options that is mostly useful and to hide those that are not needed. For example, I would choose 9-point zone (in a square) over 5-point zone (in a cross) and that spot AF is really not needed for most of the time and its AF sensitive area is just too small (for it to successfully lock on patterns). Besides, the user has to look through the viewfinder to see what is selected which is rather inconvenient. There is no indication on the body, neither on the top LCD nor on the rear (rough indication is given, though, i.e., the current AF point/zone selection method).

After all, I would rather want to have the flexibility to contract or expand the size of the zone (instead of a fixed five or nine points), say for example, via (the turning of) the rear dial, say, rather than to choose from these two fix-sized patterns.

AF Selection Accuracy (Multi-point): For stationary objects and in One-shot AF mode, it does successfully choose the closest object with some contrast within the frame correctly for most of the time. However, when the objects start moving and in AI servo mode, I found that for quite some times the camera missed the target and just chose a focus point which was contrasty enough and usually it was a stationary object within the frame. :-( So, this 61-point AF system might be not mature enough, I'm afraid. I think it will work better when the subject stands out of a background that is far behind, but this would not be practical enough. (N.B. Standard universal settings for the AF scenarios/cases was chosen, i.e., the first default option.)

AF performance and Accuracy (One-shot AF): The AF speed is lightning fast. The accuracy is acceptable with the standard non-spot AF used. With spot AF of the central AF point, it seems that the AF accuracy would decrease. As mentioned above already, it maybe owing to the fact that the AF area is just too small so that it cannot lock on enough patterns for a larger area for better light information analyses.

AF performance (AI Servo): It does track object well enough as long as the active AF points and area are manually chosen and cover the moving subject, no matter it is in the image centre or not.

Video Quality (in Full HD): I don't notice any noticeable difference in video quality with the default MPEG4/H.264 IPB encoding mode when compared to that of the K-5 (motion JPEG in 720p), when the movies are viewed on an old 1280 4:3 19" monitor. There is no continuous focus tracking available anyway and the Live View AF of the 5D3 is terribly slow by today's standard, i.e., not much improved over the 5D2. It is really rather disappointed to see that Canon has improved nothing after 3.5 years. Now the Pentax LV AF is much faster than the Canon's, with either K-r/K-5 or now the K-01. So, what Canon has been doing these years?!

Conclusion: Compared to the 5D, which was the first Full Frame DSLR body at an "affordable" price (at that time when I bought it, it was a no-brainer as Nikon was still using their DX format) and compared to the 5D2, which was the first Full HD video capable DSLR and with the highest ever resolution 135 FF sensor at its time (already equipped with gapless micro-lenses on sensor), the 5D3 is actually somehow a failure in terms of IQ, build, performance and usability. There is no major breakthrough of any of it, the new multi-point AF system is a catch-up to the Nikon competition, but it is not performing very well than originally expected. Maybe my expectation was too high.

Also as a Canon user myself (although I do not have many Canon lenses as I have for the Pentax), I am really much disappointed with the 5D2 and actually Canon. Surely the 5D3 is still a better camera body than the 5D2 overall, but the 5D3 is selling for $3500 whilst the 5D2 is only for $1800. So, one can actually buy almost two 5D2 bodies with the cost of a single 5D3. So, what's the benefits then? A more updated AF system plus slightly better high ISO performance. A new intra-frame H.264 codec for movie with a higher speed rolling shutter on sensor. That's all about it?!

Btw, I think the 5D2 is something that is the best buy right now for a FF system which means more control on (shallower DoF) plus more resolving power and a bigger view of the viewfinder for ultra-wide angle shootings, which is yet the three major meanings of FF (should also include better sensor IQ, theoretically). Besides, the high performance Canon L glass is another reason for why still Canon. USM and IS plus the highest corner to corner sharpness even near wide opened are the reasons. Nonetheless, as a Pentaxian, whether you like the taste of Canon glass is yet another story. And, the coatings of Canon L glass often suck (when it is compared to the latest generation of SMC of Pentax)! The latest Canon SWC may change that, but there are still not many L lenses have it. I've verified it with the EF24/1.4L II and finally that I agree I've found a Canon glass that could match with Pentax in terms of flare control, but still not so for the colour rendition. Besides, it is interesting to note that the name of SWC looks similar to that of SMC, with an inverted M actually (with more imagination! :-)). It seems that Canon is trying to emulate Pentax in this aspect.

After all, I'm afraid at this moment, with the introduction of the 5D3 and yet the never-marketed 1DX (as it seems to be), this would really mark the decline of the big giant Canon in the history of photographic gear, unless Canon should really do more to change the current unfavourable situation they've created for themselves, soonest, and to address at least some of the major shortcomings that I mentioned above. Not even to say they are the only major camera manufacturer that has not a mirrorless system. I believe that meanwhile the latest Nikon D800/E is surely a better FF offer by considering the body alone. So, Canon should think about why they haven't improved many things within a longish 3.5-year time-frame and have let Nikon to have the chance to get ahead of them! The laziness or reluctance of them must be blamed at the end of the day!! >:-(


Related:-

• 5D3 Vs K-5 ISO Shootout (100 to 25600)
• 5D3 Vs NEX-3 ISO Shootout (200 to 3200, Outdoor Scene)