Posted by Steve Foss on October 27, 2006 at 10:26 PM
I've been educated to believe that digital noise is dependent only on the sensor and the interplay of iso and exposure. The higher the iso, the more noise. The heavier the underexposure, the more noise. The smaller the sensor with megapixels packed in, the more noise. You know.
But I've just gotten a question from a friend who swears that the noise he gets from his Canon D-Rebel XT (350D) is more pronounced with his consumer grade 75-300 than with the 300L f4.
My guess is that, since his images from the zoom are softer and less rich than the L prime, he's sharpening more and bumping contrast and saturation, which all accentuate noise.
But here's my question. Can lens selection contribute to noise?
Interesting question Steve. I've seen nothing scientific about it, or for that matter any discussion at all. However I saw what may be something similar today. Long story short, I'm shooting an indoor job with a light kit and 100 ISO. The images (scientific illustration) need to be very sharp, so I was using my old 35-70 f/2.8 rather than my usual but slightly softer 24-120VR.
Mid-stream I was asked to get some shots of the scientists at work, so I switched to a camera mounted strobe set @ -1 for ceiling bounce (plus a green gel to match the florescent lighting and the camera set for florescent WB, if that makes any difference), then switched to 800 ISO and started the shooting. I needed something a little wider, so I put on the 24-120VR and shot some frames with that and no other changes, then back to the 35-70.
Looking at the results on my laptop at the time and again this evening on my desktop with a 24" Apple monitor, the images shot with the 24-120VR definitely were more "noisy." Again, everything else was the same as I switched lenses back and forth.
Don't know what to think about it yet, but my first guess is that since the 35-70 is so sharp and contrasty compared to the 24-120VR, sensor noise was simply less apparent. Due to my lighting technique there really were no deep featureless shadows in any of the images, but I suspect the differences I'm seeing wouldn't be so apparent there.
Good question!
Posted by Basil Greber on October 28, 2006 at 8:51 AM
Hank - did you use the same apertures and shutter speeds with both lenses? Were you shooting RAW? Are the differences still visible in RAW files?
In any case, from a physics point of view, for me it seems unlikely that the lens optics itself contributes to noise. I would guess that either differences in exposure (shutter speed, ISO?) or something that happens in post-processing (including the sharpening and saturation modifications etc. in camera) increase the noise with "cheaper" lenses.
Apertures, shutter speeds, scenes and lighting were identical. All I did was switch back and forth between the two lenses. Those shot with the 35-70 f/2.8 were more contrasty, even at a glance, and for some reason the normal 800 ISO noise was much less visible. There's still noise on very close examination, but it doesn't jump right out at you as it does with the images shot using the 24-120VR.
Capture was simultanious RAW and jpeg, and the effect is the same in both files. From a curative standpoint, it will be easier to deal with the noise postprocessing in the RAW files, I'm sure.
This contract is for a minimum of 100 hours of shooting, with more likely. I'm only 30 hours into it, so there will be additional opportunities for "stolen moments" to test this. As long as I have the large light table and lighting kit set up, along with a camera stand, it should be a nice opportunity to do so under closely controlled conditions. I'm shooting 10-12 hours a day excluding setup so it may be days before I can evaluate the results and return with a post, however.
Noise can't be generated by the lens (unless for some reason the electronics in it are not well shielded) but results from different lenses do differ in contrast, color, and bokeh so it's possible that some lenses make sensor noise more visible.
Guy
Posted by Steve Foss on October 28, 2006 at 12:49 PM
Thanks everybody. Great contributions to understanding and Hank, good perspective.
John, I do actually have glasses for long distance viewing, but it's a very weak prescription and I don't evenknow where they are anymore. Someday, though.
Basil and Guy, that's essentially what I thought. Thanks for sharing your confirming experiences.
Posted by Richard Wong on October 29, 2006 at 7:23 PM
My guess is that maybe the lesser lens produces less detailed images so the noise shows up more prominently in the detail-less areas.
Discussion
6 repliesBut I've just gotten a question from a friend who swears that the noise he gets from his Canon D-Rebel XT (350D) is more pronounced with his consumer grade 75-300 than with the 300L f4.
My guess is that, since his images from the zoom are softer and less rich than the L prime, he's sharpening more and bumping contrast and saturation, which all accentuate noise.
But here's my question. Can lens selection contribute to noise?
Mid-stream I was asked to get some shots of the scientists at work, so I switched to a camera mounted strobe set @ -1 for ceiling bounce (plus a green gel to match the florescent lighting and the camera set for florescent WB, if that makes any difference), then switched to 800 ISO and started the shooting. I needed something a little wider, so I put on the 24-120VR and shot some frames with that and no other changes, then back to the 35-70.
Looking at the results on my laptop at the time and again this evening on my desktop with a 24" Apple monitor, the images shot with the 24-120VR definitely were more "noisy." Again, everything else was the same as I switched lenses back and forth.
Don't know what to think about it yet, but my first guess is that since the 35-70 is so sharp and contrasty compared to the 24-120VR, sensor noise was simply less apparent. Due to my lighting technique there really were no deep featureless shadows in any of the images, but I suspect the differences I'm seeing wouldn't be so apparent there.
Good question!
In any case, from a physics point of view, for me it seems unlikely that the lens optics itself contributes to noise. I would guess that either differences in exposure (shutter speed, ISO?) or something that happens in post-processing (including the sharpening and saturation modifications etc. in camera) increase the noise with "cheaper" lenses.
Basil
Apertures, shutter speeds, scenes and lighting were identical. All I did was switch back and forth between the two lenses. Those shot with the 35-70 f/2.8 were more contrasty, even at a glance, and for some reason the normal 800 ISO noise was much less visible. There's still noise on very close examination, but it doesn't jump right out at you as it does with the images shot using the 24-120VR.
Capture was simultanious RAW and jpeg, and the effect is the same in both files. From a curative standpoint, it will be easier to deal with the noise postprocessing in the RAW files, I'm sure.
This contract is for a minimum of 100 hours of shooting, with more likely. I'm only 30 hours into it, so there will be additional opportunities for "stolen moments" to test this. As long as I have the large light table and lighting kit set up, along with a camera stand, it should be a nice opportunity to do so under closely controlled conditions. I'm shooting 10-12 hours a day excluding setup so it may be days before I can evaluate the results and return with a post, however.
Guy
John, I do actually have glasses for long distance viewing, but it's a very weak prescription and I don't evenknow where they are anymore. Someday, though.
Basil and Guy, that's essentially what I thought. Thanks for sharing your confirming experiences.