astrolucida
Offline
• • •
Upload & Sell: Off
|
 p.5 #14 · Canon 5D MKII and Live View | |
Not owning either the 5D II or 50D, I have had to rely on what people have reported here. Let me give you my two (euro)cents on what might be the cause of the problem.
First of all, check this page from Canon: http://www.usa.canon.com/dlc/controller?act=GetArticleAct&articleID=2027
It tells that 5D II and 50D perform peripheral illumination correction in the camera, for Canon lenses, on real time, for JPEG images (which probably means not for RAW images - has anyone tested this?). The camera relies on a database provided by Canon, which records the peripheral illumination characteristics for each lens at each aperture. The user can decide which lens data to download to the camera, with the camera being able to hold data for 40 lenses at any time and the default having 26 of the most popular ones. This might also explain why some people see a problem and some don't - there is a difference on which lenses have been loaded into the camera and whether those match the alternate lens adapter chips or not.
The second piece of evidence tells that the silent shooting mode in live view has been achieved by electronically resetting the sensor (was mentioned in this thread as a quote from Canon).
The third piece of evidence is that the problem occurs only with fast shutter speeds. Earlier in the thread there was a mention that wide apertures will cause the problem, but now I am more inclined to think it was actually the fast shutter speed only, as it is usually associated with wide apertures. Could anyone test this?
The fourth piece of evidence says that the problem occurs only with unchipped lenses or when the electronic communication between the camera and the lens is disrupted.
Now, if we combine the evidence, here's my theory: the electronic sensor clearing takes a finite amount of time and starts from the side where the shutter opens first. With the camera first communicating with a chipped lens, the timing has been adjusted to be just exactly correct when using the fastest available shutter speed (1/8000s).
However, if the communication with the lens does not happen, the electronic sensor clearing is still going on when the shutter is open at the edge of the sensor. Thus, causing the effect that we see. Or, then maybe the camera gets some important data from the lens that affects the planned timing. After all, the best that the camera can do is to trip the shutter so that the electronic clearing is completed right on schedule, just before the shutter opens up to illuminate each position of the sensor.
Canon would have never discovered this as they would have tested the delay with the fastest EF lens and shutter time and purposefully set the delay so that this combination would just barely work. The "feature" would not have been an intentional one to disable alternate lenses. This might also be the reason why Canon advices not to use silent mode with TS-E lenses (being manual focus, maybe the lens communication happens faster).
Or, the other theory is that the effect applies to all lenses but for chipped lenses the peripheral illumination control is performed. If the lens is not in the database in camera, the silent live view mode shutter vignetting is corrected in any case. Or, as the chipped lenses and adapters may mimick some Canon lenses, the camera might apply the correction, which fully corrects the live view vignetting but not completely the lens vignetting (as was seen in one of the image examples in this thread, with a trace amount of vignetting remaining in the corner of the image).
However, either purposefully or by a bug in firmware, this correction is not applied to non-chipped lenses. If this theory is true, there should be a difference even for Canon lenses between RAW and JPEG images for the live view vignetting control. If the earlier timing theory is correct, there should be no difference between RAW and JPEG images. Anyone care to test?
The reason for the vignetting is the same as described above but Canon have purposefully set the delay in shutter tripping to be shorter than the minimum, as very fast shutter speeds are very seldom used with live view, and when they are, the ISO setting is usually low (plenty of light available), so the noise is not increased perceptibly with the vignetting correction.
In any case, it simply seems to be a complicated combination of unforeseen circumstances that bring the problem out. As there is already a good workaround to handle even the seldomly occurring case, and as it does not affect Canon's own lenses, I would be inclined not to expect a correction by Canon.
As Samuli has said, typical other uses of non-chipped lenses like telescopes with adapters really seldom use speeds that are faster than 1/2000s. In fact, the only situation I can think of in astrophotography is photographing the Sun (with a proper Sun filter in front of the scope, may I remind!). However, in that case the vignetting is not a problem as the Sun is in the middle of the frame and outside there is only black sky (severely underexposed blue sky, that is) where any amount of vignetting simply disappears. And when the Sun fits only partially, then you need such focal lengths that the telescope aperture requires longer exposure times in any case.
I am not yet completely comfortable with my theories as I think that more evidence is required. If someone tests whether there is a difference between the RAW and JPEG, we might understand the situation better. Also, note that you should not use DPP for processing the RAWs, as it perfoms the same type of corrections as the camera does. Or, you could try whether a RAW taken by a non-chipped lens would actually be corrected by DPP. For the other test, use another RAW converter and disable any vignetting or lens-related corrections.
|
