p.1 #1 · Voigtlander WA on NEX-5N colorshift testing
So, as I had volunteered on the NEX Images thread, here are a series of test of vignetting and color shift of the Voigtlander 15/4.5 LTM on the NEX-5N. I believe Jacob D is going to try to do the same tests with the CV 12/5.6.
Quick recap of the methodology: I shot outside into a sunny sky with a white tissue paper stretched over the lens. ISO100, exposure comp at +2EV, aperture priority. I shot in raw, then in Lightroom, I white balanced using the center of the image, and then increased exposure to just before clipping. Here are the shots:
I have to say, I'm surprised that this test showed as strong vignetting/color shift as it did. In actual practice of having shot this lens/camera combination for a couple of weeks, the problem rarely rears its head.
p.1 #5 · Voigtlander WA on NEX-5N colorshift testing
Very awesome thread, alwang. How this particular lens exactly performed on the 5N was still in question (it was assumed that it worked good in non-test shots by people who have owned it). Good to see something conclusive.
p.1 #7 · Voigtlander WA on NEX-5N colorshift testing
ken.vs.ryu wrote:
do you have the 16mm E-mount? Would love to see what these 2 produce with real world shots.
Sorry, I don't have the 16mm. But the methodology is straightforward enough (as suggested by douglas and jacob), that I'd just encourage others to give it a try and post the results. We might have to rename the thread
p.1 #9 · Voigtlander WA on NEX-5N colorshift testing
Sosua wrote:
Good stuff - I only notice it very occasionally in the real world and its likely invisible to most people.
I think the 12mm F5.6 is 'easier' in this regard? although have not got the lens.
Only issue I can really see is that the "most people" who would find it invisible aren't the same people who would want to use manual focus RF glass in the first place.
Having the M mount version of the 15 on the original NEX5, I can see a marked improvement with the 5n, that should process away in most instances, though again, if your looking for color cast you typically will see it, which I think is true of most any optical defect with a given lens/body combo.
Hard to beat a $150 16mm f2.8 E mount and the $65 UWA converter for bang for the buck though. Older non M 15mm f4.5 might not be a bad price, and even the M mount is reasonable i guess, but when getting into the 12mm f5.6 at $850, I just didn't find the performance there, nor really enough use of the focal length to justify what I paid for it.
p.1 #10 · Voigtlander WA on NEX-5N colorshift testing
alwang wrote:
So, as I had volunteered on the NEX Images thread, here are a series of test of vignetting and color shift of the Voigtlander 15/4.5 LTM on the NEX-5N. I believe Jacob D is going to try to do the same tests with the CV 12/5.6.
Quick recap of the methodology: I shot outside into a sunny sky with a white tissue paper stretched over the lens. ISO100, exposure comp at +2EV, aperture priority. I shot in raw, then in Lightroom, I white balanced using the center of the image, and then increased exposure to just before clipping. Here are the shots:
...
I have to say, I'm surprised that this test showed as strong vignetting/color shift as it did. In actual practice of having shot this lens/camera combination for a couple of weeks, the problem rarely rears its head.
Here is the same with the CV 12/5.6. I had a helluva time with this because it kept picking up subtle textures in whatever I covered the lens with (hints of the same in the f16 example above with the CV12). In the end I found a piece of 1/4" UHMW (white, semi-opaque plastic) that I used.
I edited this with change in method of generating the profile. Now shooting in aperture priority, ISO 100, +2 EV exposure comp, focus at infinity. Only adjustment in LR is WB at approx center of image.
p.1 #13 · Voigtlander WA on NEX-5N colorshift testing
It must be different processing. White balance is probably the key, I know we both balanced off the center, so I'm not sure why my results are more vs. purple. My WB values were 5100 and 17.
I tried lowering my exposure then WB again, but the results didn't change much.
I know in practice that the shift is usually more purple, but I guess this still shows the potential shift. For creating a cornerfix profile the actual color balance is said not to matter, this is the sort of thing that makes me wonder. If our shots are near-clipping at the center and the corner color varies from yours' to mine, that seems like the sort of thing that would matter in creating profiles.
p.1 #16 · Voigtlander WA on NEX-5N colorshift testing
No problems.
Focused at infinity, yes.
Al and I talked about this a little before doing it, but my method was just slightly different. I wonder if it makes a difference one way or the other...
Rather than using exposure comp, I used manual mode, exposed to the far right but short of clipping (using live histogram). Adjust the exposure as needed in LR to bring it as close as possible to clipping without doing so. (less than 1/3 stop push in LR)
p.1 #17 · Voigtlander WA on NEX-5N colorshift testing
I believe that +2EV is recommended by most that use CornerFix, although we can choose whatever we want for our own comparisons. However, if were going to compare multiple lenses against each other, we should keep it consistent, because the more exposure, the less vignetting and color shift will be visible, so it could skew the comparisons.
You could still use M mode, just meter for +2EV to keep everything on the same playing field. Thanks!
edit* I just realized alwang also raised exposure after the fact in the converter, so were going to have a tough time comparing appropriately. We really need to do these over by using the same exposure and not pushing it in the converter, if we're going to do comparisons.
p.1 #18 · Voigtlander WA on NEX-5N colorshift testing
Yes, I was focused at infinity as well. I could certainly post the same images without the Lightroom exposure increase... (so it would be just the +2EV exposure comp when shooting)
p.1 #20 · Voigtlander WA on NEX-5N colorshift testing
So I've updated my samples based on Douglas comments.
To be on the same playing field as Alwang, I shot in aperture priority, +2 EV exposure comp, ISO 100, focus at infinity. The only LR adjustment was WB from approx center.
As a reference value, the luminance at the approx center of the frame is 84%.
...
Now thoughts that I have... the first time I shot these samples I exposed manually close to the clipping limit and pushed them a little in LR to get closer (less than 1/3 stop in all cases). In that scenario much less shift is seen. Does it make sense to create profiles at various exposure levels to suit the exposure of the scene? Should I have a left, center, and right (in reference to the histogram) profile for each lens? These are probably questions for the creator of Cornerfix, but I think out of curiosity I will make those profiles, then see how they work on images where the histogram is biased towards one end or the other.
