Nikon 18-300mm image softness
Discussion
I've just got myself one of these from MPB for my D7100 and I'm disappointed with the image softness at wide aperture. Not sure if it's my skills, expectations or a dud lens.
Things are much better from f8 upwards, but at f3,5-6 (even at 18mm) the image is soft all over. Doesn't look like a focus issue, as depth of field at 18mm shouldn't be a problem, surely. Will try and post some 1:1 crops shortly.
Anyone else using this lens?
Things are much better from f8 upwards, but at f3,5-6 (even at 18mm) the image is soft all over. Doesn't look like a focus issue, as depth of field at 18mm shouldn't be a problem, surely. Will try and post some 1:1 crops shortly.
Anyone else using this lens?
Thanks all.
I've double-checked some of my test shots, and I reckon the problem is only at the widest end of the lens. The test shots at 300mm f6.3 were sharper than the f22 one. (I'd had some earlier blurry results when zoomed, but I think I can put those down to too much faith in AF and VR, rather than using a tripod...)
Looking back at the bad shots I've taken, they're all at 18mm, focussed about 5-10m away, and at 18mm I should have a huge DOF even at f3.5.
Is it possible that the focus mechanism is hitting (or missing) and endstop somewhere?
I've double-checked some of my test shots, and I reckon the problem is only at the widest end of the lens. The test shots at 300mm f6.3 were sharper than the f22 one. (I'd had some earlier blurry results when zoomed, but I think I can put those down to too much faith in AF and VR, rather than using a tripod...)
Looking back at the bad shots I've taken, they're all at 18mm, focussed about 5-10m away, and at 18mm I should have a huge DOF even at f3.5.
Is it possible that the focus mechanism is hitting (or missing) and endstop somewhere?
Which object is your single spot focus fixed on according to the viewfinder?
What happens if you switch to manual focus and then use the focus ring and mark 1 eyeball?
What happens if you place a cereal packet 60cm in front of the camera in autofocus mode and focus on the picture and text on the front at 18mm?
Do you see any differences when adjusting the autofocus fine tuning settings on the D7100?
What happens if you switch to manual focus and then use the focus ring and mark 1 eyeball?
What happens if you place a cereal packet 60cm in front of the camera in autofocus mode and focus on the picture and text on the front at 18mm?
Do you see any differences when adjusting the autofocus fine tuning settings on the D7100?
eltawater said:
Which object is your single spot focus fixed on according to the viewfinder?
Focus was on the brickwork. But it seems irrelevant, as at 18mm/f3.5 at around 10m to subject my DOF should be many, many meters. Also in the full shot there's grass/gravel from about 3m to 30m. It's all equally soft.
eltawater said:
What happens if you switch to manual focus and then use the focus ring and mark 1 eyeball?
What happens if you place a cereal packet 60cm in front of the camera in autofocus mode and focus on the picture and text on the front at 18mm?
Do you see any differences when adjusting the autofocus fine tuning settings on the D7100?
Cheers. I'll give those a shot tomorrow. My gut says it's not an autofocus issue as the test shots of the same scene at longer focal lengths (where accurate focus is more critical) all seem pin-sharp.What happens if you place a cereal packet 60cm in front of the camera in autofocus mode and focus on the picture and text on the front at 18mm?
Do you see any differences when adjusting the autofocus fine tuning settings on the D7100?
Has anyone used the 'focal' calibration software Is it worth the money?
Edited by silentbrown on Sunday 12th April 23:34
Try focusing on the pot instead of the brickwork. Once focused, switch focus to manual before taking the shot otherwise there's a risk you'll get accidental auto refocussing when you press the shutter.
I'm assuming you've turned off the vr when on the tripod.
The f3.5 shot looks like it's just generally out of focus. Going to f22 and seeing things back into focus suggests the focus point is not where you think it is and that by narrowing the aperture you're keeping the same focus point but increasing the depth of field to the point where those areas are now in focus.
Sharpness at the long end may be a bit of a red herring as you tend to zoom in on one prominent subject and so give the autofocus less ambiguity on what to focus on.
I'm assuming you've turned off the vr when on the tripod.
The f3.5 shot looks like it's just generally out of focus. Going to f22 and seeing things back into focus suggests the focus point is not where you think it is and that by narrowing the aperture you're keeping the same focus point but increasing the depth of field to the point where those areas are now in focus.
Sharpness at the long end may be a bit of a red herring as you tend to zoom in on one prominent subject and so give the autofocus less ambiguity on what to focus on.
Thanks all. Some very good suggestions, and I'm a lot closer to working out the problem, which does appear to be AF related.
Looking at the EXIF data with EXIFTool, the recorded focus distance is under 3m. The actual distance is 10m, which explains a lot!
The camera thinks it's got good focus. Focus is quick, lots of AF points show on detail sections of the shot
Manual focus is trickier, because the image still looks sharp in the viewfinder when zoomed to 18mm. Zooming in, focusing then zooming out to take the shot gives a sharp image but I'm wary of this! I haven't tried using liveview for manual focusing, but I expect that would work best.
I switched to my 18-55 lens, and the image is a lot sharper using AF- but the recorded focus distance is still way out, at 4m. Looking back at EXIF data of landscapes I've shot with that lens they all have a closer focus distance than expected, but are typically shot around f8 so everything's sharp anyway...
I'm going to try experimenting with AF Calibration, to see if it makes a difference. The odd things is that it now seems like a body issue rather than a lens one.
Looking at the EXIF data with EXIFTool, the recorded focus distance is under 3m. The actual distance is 10m, which explains a lot!
The camera thinks it's got good focus. Focus is quick, lots of AF points show on detail sections of the shot
Manual focus is trickier, because the image still looks sharp in the viewfinder when zoomed to 18mm. Zooming in, focusing then zooming out to take the shot gives a sharp image but I'm wary of this! I haven't tried using liveview for manual focusing, but I expect that would work best.
I switched to my 18-55 lens, and the image is a lot sharper using AF- but the recorded focus distance is still way out, at 4m. Looking back at EXIF data of landscapes I've shot with that lens they all have a closer focus distance than expected, but are typically shot around f8 so everything's sharp anyway...
I'm going to try experimenting with AF Calibration, to see if it makes a difference. The odd things is that it now seems like a body issue rather than a lens one.
silentbrown said:
Has anyone used the 'focal' calibration software Is it worth the money?
I use it, it is good but if your issue isn't an autofocus problem then it won't help. Also depending on your camera you may only be able to store one AF fine tuning profile for each lens - if you run FoCal and work out a profile for 18mm then that may not be appropriate for 300mm so could make things worse at the long end.I've never used the 18-300 but I've used other superzooms and never been impressed. Any lens design, even high-end primes, involve a number of compromises. Zoom lenses involve more compromises because the focal length obviously is not fixed. Superzooms are compromised even further as they have to try and achieve acceptable sharpness across a huge range of focal lengths. Superzooms are usually aimed at the low end of the market, i.e. one lens to do it all so are priced at a low price point which just compounds the issue further. I realise that this won't help you much but having been caught out with this issue myself in the past I tend not to expect anything great from a superzoom.
Simpo Two said:
Yep. If you're getting particular about IQ then you're not really starting form a good place.
Understood. I know there's always a compromise with any zoom, but the initial results with this were way below expectations. It's definitely an AF problem, rather than an inherent lens sharpness issue. Images manually focused with liveview are perfectly sharp, even at widest aperture.
I've done the dance of the angled ruler, and it confirmed what I was seeing in the EXIF data. The lens was focusing way too near. I didn't notice this in my original shots because the foreground of my shots was still too far away to be in focus!
After some back-and-forth I've tweaked the AF fine adjustment for the lens to +10 and that's a huge improvement. Not sure if I need to get this more accurate at this stage.
silentbrown said:
It's definitely an AF problem, rather than an inherent lens sharpness issue. Images manually focused with liveview are perfectly sharp, even at widest aperture.
I've done the dance of the angled ruler, and it confirmed what I was seeing in the EXIF data. The lens was focusing way too near. I didn't notice this in my original shots because the foreground of my shots was still too far away to be in focus!
After some back-and-forth I've tweaked the AF fine adjustment for the lens to +10 and that's a huge improvement. Not sure if I need to get this more accurate at this stage.
OK, that sounds like definite progress, well done. I believe there are places that can calibrate bodies and lenses together, but now might be the time to ponder replacing the 18-300 with something better, eg two or even three lenses.I've done the dance of the angled ruler, and it confirmed what I was seeing in the EXIF data. The lens was focusing way too near. I didn't notice this in my original shots because the foreground of my shots was still too far away to be in focus!
After some back-and-forth I've tweaked the AF fine adjustment for the lens to +10 and that's a huge improvement. Not sure if I need to get this more accurate at this stage.
Simpo Two said:
OK, that sounds like definite progress, well done. I believe there are places that can calibrate bodies and lenses together, but now might be the time to ponder replacing the 18-300 with something better, eg two or even three lenses.
I've only had it three days! 
It gets curiouser, though. I'm getting good results overall, but the occasional shot is just rubbish quality, and it seems unpredictable.
A few on Flickr here. https://flic.kr/s/aHsmMyXRNn
The first three shots are fine. The second landscape has identical aperture, shutter speed (and, according to exif, a focus distance of 8 metres) as the first landscape - but as soon as you zoom in all the detail is shot. I've added crops of the two landscapes for comparison.
My first thought was camera shake - but even hand-held, a shutter speed of 1/160s at 18mm with a VR lens shouldn't look like that.
You can still quite easily get camera shake at 160/s even with VR, it's not completely fool proof. The distortion around the trees closet to you on the left could easily be normal lens distortion at wider apertures so I'd not get too hung about that. The lack of detail in the path leans me more towards being slightly out of focus or slight movement as there's no obvious candidate that's sharp enough for a focus point.
Next time try 320/s and turn off the VR to rule out interference from that. You can also attach it to a sturdy tripod and take multiple shots to compare.
Next time try 320/s and turn off the VR to rule out interference from that. You can also attach it to a sturdy tripod and take multiple shots to compare.
eltawater said:
Next time try 320/s and turn off the VR to rule out interference from that. You can also attach it to a sturdy tripod and take multiple shots to compare.
I'll let you know how it goes. I've never had any shake problems with wide lenses before. I just added a cropped handheld shot to the album with the same lens at full zoom (1/200s) and that's pretty solid. The domes are 18 miles away... My rule of thumb with 35mm was to use the same fraction of a second as the focal length, (or half that if you're feeling lucky). So 18mm would mean 1/18s or faster. Not sure if that still applies in a digital APS-C world!
Gassing Station | Photography & Video | Top of Page | What's New | My Stuff