rscheffler wrote:
Front element/entrance pupil only needs to be 25mm for a 35/1.4 lens. Consider the many 35/1.4 lenses available for the Leica M rangefinder system and they're all much smaller than mainstream mirrorless/DSLR 35/1.4s. For example the Leica 35/1.4-M is 46mm filter size.
A reason for the huge optics of modern fast lenses (both in diameter and in groups/elements count) is to more easily improve image quality/sharpness. Look at Zeiss Otus, Sigma Art (particularly their massive 35/1.2 for mirrorless), Canon's EF 35/1.4 v2... The smaller the elements and the fewer there are, the more 'work' each element has to do to form the desired image. And the shape of each element, the spacing, the assembly all need to be much more precise.
With the short mirrorless flange distance, an important consideration is that light rays reach the entire sensor area as perpendicular to the sensor surface as possible. Hence the reason many mirrorless lenses have large rear elements to minimize the strongly oblique light rays typically projected from small diameter rear elements that sit near the sensor surface. Strongly angled light travels farther through the glass sandwich that covers the sensor itself, resulting in additional image refraction that can degrade image quality if it is not accounted for in the lens design. Such light rays also require more help from customized microlens arrays above the sensor's pixel wells to redirect the light as optimally as possible down into the pixel wells to the actual light sensitive photosite....Show more →
Greatly appreciate the detailed explanation and insight.
So far, at least based on my personal experiences/observations the RF 35, Sony GM 35 (sold), and Sigma 35/1.2 (still own):
The GM and RF feel very similar with regard to size/weight in hand.
The GM is better corrected for distortion and vignetting.
AF in stills so far, seemed better (more consistently accurate and without anomalies) with the Sony options. But that could very well just be a difference in body and/or user. Will be using the RF a lot more in the coming weeks, so my mind may change once I figure out how it works best.
Didn't love the output of the GM though.
Output from the RF has been both good and bad at times.
Sigma beats both the RF and GM with regard to general look/output for my personal taste. Yes, it is huge and heavy in comparison. lol.
I tried the 35mm f/1.8 IS Macro, and just didn't like the STM focusing system. I returned it and placed an order for the RF 35mm f/1.4 VCM. I am OK with the profile corrections as long as the final output is good, which appears to be the case from the samples I have seen. Looking forward to having a fast 35 again.
Playing with the files in LR a bit, and just figured I'd get some figures myself with regard to the manual distortion correction and resolution loss.
Seems to need about a +30 to 35 distortion correction to mimic what the profile does for most images. After constraining crop, seems like Sam Hurd's findings were pretty accurate. Its essentially just over 1 megapixel of resolution cut off the sides in total. Not a huge deal, especially if you're using a body with a higher resolution sensor. Still annoying though.
For many shots, I'm perfectly fine leaving the distortion as is.
artsupreme wrote:
Same here, I'm finding myself unchecking the lens correction box very often with this lens.
Yep. Most of the shots where I can't just call it good are ones where there are some prominent lines that look way off. Otherwise, I like the distortion, as I admittedly love imperfections in lenses that give them an interesting look.
On another note, I was mistaken in my original measurement quote for the square metal hood I'm using. It's actually about a 2 inch savings off the front of the lens, and it has yet to induce any weird optical aberrations/vignetting or flaring. Thinking that I can safely recommend this to anyone looking to shorten the length of the package by a couple of inches (unless you don't use hoods).
Shot with the Sigma 35/1.2 because it seemed fitting. DSC01829 by John Dizzo, on Flickr
This is what the distortion looks like when it's not corrected with the LR profile. With the correction, all lines are straight, of course, but the left side of her train was cut off in the straightening process, so I decided to go without it.
96whiteknight wrote:
This is what the distortion looks like when it's not corrected with the LR profile. With the correction, all lines are straight, of course, but the left side of her train was cut off in the straightening process, so I decided to go without it.
Lovely picture, but is the focus plane doing weird things, or is it just me? Like why is the back brick wall soft in the middle, but sharpening up towards the edges? I mean, it’s still soft, as intended, but it’s less soft as you go further from the center. Or am I crazy (entirely possible).
comotionfilms wrote:
Lovely picture, but is the focus plane doing weird things, or is it just me? Like why is the back brick wall soft in the middle, but sharpening up towards the edges? I mean, it’s still soft, as intended, but it’s less soft as you go further from the center. Or am I crazy (entirely possible).
This is seen in many wider angle lenses, especially for the Leica M rangefinder system, in part because these designs are typically optimized for small size, which causes a stronger combination of mechanical and optical vignetting. It effectively means that light from the edges of the image area is passing through a smaller, truncated maximum 'aperture' in the lens (which is seen in images as off-center 'cat's eye' bokeh ball shapes), which produces the same result as stopping down and becomes progressively stronger away from the image center as 'cat's eye' bokeh becomes more pronounced (at image center, light passes through the full maximum aperture).
As we've already seen, the 35 VCM has pretty strong cat's eye bokeh in the image periphery due to its fairly high optical/mechanical vignetting, likely a compromise caused by its relatively compact design (though is still unfortunate). In the uncorrected (for distortion) image above, you're seeing farther into the edges/corners than Canon's lens designers intended, where the effect is even stronger.
"In the center of the frame almost every lens will render a perfect (bokeh) circle, but only lenses with very low optical vignetting will keep this shape in the corners, as optical vignetting leads to the truncation of light circles towards the borders of the frame. This also leads to the depth of field increasing towards the corners, so most lenses will show a higher amout of blur in the center of the frame compared to the corners."
Another possible factor is unfavorable field curvature. With some lenses the plane of focus bends more towards infinity in the image periphery area, which results in background areas in the image periphery looking sharper than in the image center. But I think with the 35 VCM it's probably mostly due to relatively high optical vignetting.
The 35 VCM's MTF curves also suggest that the image periphery will have a more nervous look due to increasing divergence of the sagittal and tangential plots as they move off-center. And I think we can kind of see that in the background brickwork in the image above.
The 50 VCM might be the 'sleeper' lens from the trio in respect to smoothest off-axis character/rendering and focus transition. It still appears to have high optical vignetting as seen in the first sample image here.
But the background looks good in the second image where there is less subject/background separation (though still difficult to really tell at such low resolution).
IMO it looks better than the RF 50/1.2L's MTFs (below), particularly outside the 10mm image circle area where the sagittal and tangential plots track very closely together:
I generally like the balance of compromises in these VCM primes but conversely I'm still somewhat disappointed that so many of Canon's RF lenses have fairly high optical vignetting (resulting in strong cat's eye bokeh), whereas some of the competition's lens designs appear to place higher priority on minimizing this specific characteristic.
rscheffler wrote:
This is seen in many wider angle lenses, especially for the Leica M rangefinder system, in part because these designs are typically optimized for small size, which causes a stronger combination of mechanical and optical vignetting. It effectively means that light from the edges of the image area is passing through a smaller, truncated maximum 'aperture' in the lens (which is seen in images as off-center 'cat's eye' bokeh ball shapes), which produces the same result as stopping down and becomes progressively stronger away from the image center as 'cat's eye' bokeh becomes more pronounced (at image center, light passes through the full maximum aperture).
As we've already seen, the 35 VCM has pretty strong cat's eye bokeh in the image periphery due to its fairly high optical/mechanical vignetting, likely a compromise caused by its relatively compact design (though is still unfortunate). In the uncorrected (for distortion) image above, you're seeing farther into the edges/corners than Canon's lens designers intended, where the effect is even stronger.
"In the center of the frame almost every lens will render a perfect (bokeh) circle, but only lenses with very low optical vignetting will keep this shape in the corners, as optical vignetting leads to the truncation of light circles towards the borders of the frame. This also leads to the depth of field increasing towards the corners, so most lenses will show a higher amout of blur in the center of the frame compared to the corners."
Another possible factor is unfavorable field curvature. With some lenses the plane of focus bends more towards infinity in the image periphery area, which results in background areas in the image periphery looking sharper than in the image center. But I think with the 35 VCM it's probably mostly due to relatively high optical vignetting.
The 35 VCM's MTF curves also suggest that the image periphery will have a more nervous look due to increasing divergence of the sagittal and tangential plots as they move off-center. And I think we can kind of see that in the background brickwork in the image above.
The 50 VCM might be the 'sleeper' lens from the trio in respect to smoothest off-axis character/rendering and focus transition. It still appears to have high optical vignetting as seen in the first sample image here.
But the background looks good in the second image where there is less subject/background separation (though still difficult to really tell at such low resolution).
IMO it looks better than the RF 50/1.2L's MTFs (below), particularly outside the 10mm image circle area where the sagittal and tangential plots track very closely together:
I generally like the balance of compromises in these VCM primes but conversely I'm still somewhat disappointed that so many of Canon's RF lenses have fairly high optical vignetting (resulting in strong cat's eye bokeh), whereas some of the competition's lens designs appear to place higher priority on minimizing this specific characteristic. ...Show more →
Still been playing with this lens when I can. Just been tough, as I picked up a few other lenses at the same time. I will say though, that in certain conditions, I really do like the uncorrected look this lens produces. Couple of shots from the weekend.
