Help with DoF width-wise

[leansoli]

Edit: Sorry for the lenght, didn't realize I made such a long post after I hit the submit button! And forgive any typo, English is not my main language

Hi guys!

I must admit I learned a lot with the tutorials on this page, and reading them acted as some sort of catalyst to buy my first digital SLR. That happened a couple of years ago and by reading here along with playing with the camera enabled me to take some shots I never thought I would be able to get. So, kudos to the people that make this site and keep the good work!


Regarding the purpose of this thread, this last sunday I had my cousin's baptism. I am used to take pictures for my family and/or friends, non-professional pictures but perhaps above average pics. But this time I couldn't get a decent shot.

I was reading this thread and I could understand the problem. But in my case, I needed a neat DoF throughout the width of the picture rather than a background-foreground case.

I have uploaded the picture I speak about here.

There you can see that the guys standing on the far left and the far right are not as sharp as the people in the center. I estimate I was 10 meters away from the group. I estimate that the distance from the baby in the middle to the teenager right above the baby (but in the background) is around 3 meters, and I have no idea what could the distance be between the guys on both sides.

I am happy with the sharpness I got in the center of the picture (notice the palm tree reflected in the glass of the door behind all the people. That door could easily be 5 meters further away, adding a total of 8 meters, probably 10mts from the baby to that door.). But I am really disappointed about the sharpness I got on the sides of the shot

In this case, the distance from camera to baby, compared to the distance from camera to either guy on the side, is almost the same. Why would I lose so much sharpness if the difference in the distances I speak about is negligible?

In other words, background to foreground: I get decent sharpness in like 10 meters. Left to Right, I lost half my sharpness in like 6 meters?

Technical stuff that may help you help me:
Consider this is a Jpeg for uploading purposes. The original is a RAW file, resolution of 4272*2848 and RAW file is brighter than this JPG (had to tune the brightness a bit).
I own a Canon 450D, this shot was taken at f3,5 1/80s ISO 100 on a 28/80 Tamron lens @ 28mm. I also used a flash that acted as a filler since baby's mom wanted the pic to be taken like that. The sun was in front of me, but the church casted that sort of shade...

Thanks in advance!

[Peter Ryan]

Hi Leansoli and welcome to CiC.

We tend to like to address members by their preferred name and you can include this on the left by putting you name in your profile where it asks for Real Name.

I will have a go at this and I am sure others will also have more to contribute. Thanks also for providing all the shooting information.

From my understanding DOF should be the same for all subjects at the same focal distance from the camera when the camera focuses at that point; in this case being the front row of the group. Lenses are sharpest in the middle of the frame and also in the middle of their zoom range. Sharpness can fall off at the edges when you use the lens at it extremes, particularly wide angle lenses.

The only reason I can think of is that you were using your lens at the widest possible focal length (28mm) and that your lens has soft focusing at the outer edge of the image frame at the extreme ends of the zoom range.

Lenses also work best around 2 stops down from wide open. In this case you appear to be working with the aperture wide open at f3.5. The combination of working at the extreme end of the focal length range and wide open aperture compounds the issues.

I would suggest operating around f/8 and increasing your ISO to 200 or even 400 if required to hand hold the shot (assuming no tripod) and see if the results improve but I feel your answer lies more in the lens than DOF.

[jeeperman]

I am a bit of a rookie here myself but from what I can see....as you say you have 4 people deep in focus. Those people are standing close together front to back. If you look at your first row and count 4 people out to left or right you begin to lose focus. Those people are shoulder to shoulder.....same # of people at almost double the distance. For such a wide group you may have done better to go with a narrower opening maybe f9 or beyond. My guesstimate on fstop may be off a little, I am sure I will be corrected. Hope I have helped.


I need to learn to type faster! lol I would go with what Peter said. =)

[pwnage101]

Ideally, rectilinear lenses retain focus along a FOCAL PLANE that is perpendicular to the direction of the camera.

If you shoot such that all subjects are along a plane perpendicular to the direction of the camera, then all subjects will be in focus.

As seen on anstendig.com:

Figure 4: full body in focus! yay!


Figure 3: lower body out of focus


Objects along the FP(focal plane) in the corner of the image are always farther from the camera than objects along the FP in the center of the frame. This shows that it is possible for different objects at different distances from the camera to be in focus.
Your problem is NOT a focus problem. It is a severe lens defect

[revi]

Seeing that the door in the background seems to be in focus, you might have put the focus point too far towards the back of the group (or even after the group, on the door) . Also, you seem to have turned slightly to your right, aggravating the problem for the people on the left.

And as Peter said, using the lens at it widest aperture usually isn't optimal for such pictures, and going to 400 ISO shouldn't cause any problems, even for A4-sized prints.

[Peter Ryan]

I think Remco has added another clue here (good point Remco).

Looking at the image again, if you just pointed the camera at the crowd and focused then the focal plane is most likely behind the front row narrowing the area of perceived sharpness (circle of confusion) of the people at the edges of the front row, as they are further from the camera than the people in the middle front row and exacerbating the problem.

Do you remember where you actually focused in the image?

[Colin Southern]

I think you're all wrong

Let's do the maths ... Depth of Field for a 28mm lens on a 1.6x crop-factor camera @ 10m @ F3.5 is ...

... 55.921 metres (or from 5.444m to 61.365m), so it's not a DoF issue.

Still don't believe me?

The distance to the folks in the front row at the sides is probably similar to those in the back row in the middle ... and yet the ones in the middle at all rows are in focus, but the ones at roughly the same distance from the camera on the outside isn't - so again, not a DoF issue.

So what is it? My guess is just poor lens performance. For this reason I'd probably have shot it at around F8 @ ISO 400 (just to get the lens more in the sweetspot), but to me it's looking like a lens issue.

What lens is it by the way?

[Peter Ryan]

I think you're all wrong

- so again, not a DoF issue.

So what is it? My guess is just poor lens performance. For this reason I'd probably have shot it at around F8 @ ISO 400 (just to get the lens more in the sweetspot), but to me it's looking like a lens issue.

What lens is it by the way?

Isn't this what I said in the first post but without the maths?

[Colin Southern]

Isn't this what I said in the first post but without the maths?

Oops, I'm owned by Peter

Sorry Peter - quite right! I think I "sped-read" through the other posts a bit quickly!

[pwnage101]

I think you're all wrong

Me too me too! I'm not wrong either!

Your problem is NOT a focus problem. It is a severe lens defect

[leansoli]

To begin with, let me thank you all for taking the time in analyzing my situation and providing feedback!
Having added my real name and geographic location, I move onto the answers:

Lenses also work best around 2 stops down from wide open ... The combination of working at the extreme end of the focal length range and wide open aperture compounds the issues.
I would suggest operating around f/8 and increasing your ISO to 200 or even 400 if required to hand hold the shot (assuming no tripod) and see if the results improve but I feel your answer lies more in the lens than DOF.

I tend to take my shots with the lowest ISO possible. Tbh, I hardly use ISO 200, so reading about using ISO 400 makes me chill. I have a few prints (40*35 centimeters or 16*14 inches), but of course those prints belong to ISO 100 shots so I really have no idea how a ISO 400 shot looks like printed. I'll take a few shots at ISO 400 and have them printed... perhaps this will make me feel more confident about using higher ISOs.

...My guesstimate on fstop may be off a little, I am sure I will be corrected

I used the biggest aperture available to get as much sharpness as possible, yet the previous post also suggests to use lower apertures :/

...Also, you seem to have turned slightly to your right, aggravating the problem for the people on the left.
And as Peter said, using the lens at it widest aperture usually isn't optimal for such pictures, and going to 400 ISO shouldn't cause any problems, even for A4-sized prints.

I am right-handed and I look through the camera 'window' ('viewfinder' rings a bell for me. Is that the word? My bad :$) with my right eye, so it's totally possible that I did what you say. Next time I'll pay more attention to this and see if I tend to do it. I do recall that the sun would hit the lens, so I had to bend over a litle that day.

Again the ISO 400 is brought up. I'll definitely use it and get some prints to test it. I'll check on noise reduction later, perhaps my conception of ISO 400 and above is totally wrong.

Looking at the image again, if you just pointed the camera at the crowd and focused then the focal plane is most likely behind the front row narrowing the area of perceived sharpness (circle of confusion) of the people at the edges of the front row, as they are further from the camera than the people in the middle front row and exacerbating the problem.

Do you remember where you actually focused in the image?

I believe I used 'Evaluative Metering' and focused on the girl who is holding the baby. I believe she has a middle-grey-ish skin :P
I am staying at my hometown where I don't have my desktop computer. If I had Canon's software I could tell you precisely the spots that were in focus.

The distance to the folks in the front row at the sides is probably similar to those in the back row in the middle ... and yet the ones in the middle at all rows are in focus, but the ones at roughly the same distance from the camera on the outside isn't - so again, not a DoF issue.

So what is it? My guess is just poor lens performance. For this reason I'd probably have shot it at around F8 @ ISO 400 (just to get the lens more in the sweetspot), but to me it's looking like a lens issue.

What lens is it by the way?

The distance relations you mention is exactly what made me come here and get this thread going. In my OP I mention the difference in those distances to be negligible.

Lens is a Tamron 28/80. I know it's far from being the best, but it was a bundle I saw on ebay that matched my budget :/
I'll get a few shots this coming weekend and check their sharpness on the sides. If it is indeed a lens performance issue, what lens would you suggest as a replacement?

Thanks again for the answers!

Edit: I ALWAYS have a UV filter. Could the filter be the cause?

[Dave Humphries]

Hi Leandro,

Welcome to the CiC forums from me, good to have you join us, although ideally under better circumstances would have been preferable.

~ so reading about using ISO 400 makes me chill. I have a few prints (40*35 centimeters or 16*14 inches), but of course those prints belong to ISO 100 shots so I really have no idea how a ISO 400 shot looks like printed. I'll take a few shots at ISO 400 and have them printed... perhaps this will make me feel more confident about using higher ISOs.

You may need to apply some noise reduction to iso 400 and above if viewing/printing large, especially if cropping down from the full size, but that is far, far easier and less damaging than this kind of effect.

Iso 400 definitely shouldn't "give you a chill"; I regularly shoot at 800 - 2000, or higher.

I used the biggest aperture available to get as much sharpness as possible, yet the previous post also suggests to use lower apertures

Is it possible you have mis-read "bigger" somewhere and applied it to the diameter of the hole instead of the f/ number?
Wide open on 95% of lenses will not produce the sharpest overall image, as many have said - it is a major 'feature' of lenses that the centre is sharper than the edges wide open. Whereas at f/16 or f/22, whatever the minimum diameter is, they are generally most evenly sharp, HOWEVER not as good as a mid-range stop like f/8, which usually balances good/better centre sharpness with acceptable edge sharpness.

I ALWAYS have a UV filter. Could the filter be the cause?

No

[Colin Southern]

Hi Leandro,

To begin with, let me thank you all for taking the time in analyzing my situation and providing feedback!

No worries - it's what we do best here

Having added my real name and geographic location, I move onto the answers:

Thanks for that. It's always nicer to be able to say something like "Hi John" rather than "Hi ForceToken6697" (except in this case your name is Leandro, so calling you John would just be silly!)

I tend to take my shots with the lowest ISO possible. Tbh, I hardly use ISO 200, so reading about using ISO 400 makes me chill. I have a few prints (40*35 centimeters or 16*14 inches), but of course those prints belong to ISO 100 shots so I really have no idea how a ISO 400 shot looks like printed. I'll take a few shots at ISO 400 and have them printed... perhaps this will make me feel more confident about using higher ISOs.

For landscape - shot from a tripod - this is almost mandatory. In situations where camera shake and/or subject motion may degrade an image then it's not so clear cut ...

Everything in photography is a compromise; too low a shutterspeed and you risk your image being degraded by camera shake or subject motion - too wide an aperture (low "F" number) and you risk your image being degraded by an insufficient depth of field or operating the lens in an area where it's performance is "less than ideal" - too high an ISO and you risk your image being degraded by noise and/or having issues with the reduced dynamic range (a topic for another day), so we need to work out what's the best combination.

In your case, you need to keep the shutterspeed high enough to freeze any inadvertant movement, and minimise any camera shake ... so personally I'd be aiming for at least 1/60th at this focal length. In terms of aperture, it's generally accepted that the "sweet spot" for most lenses is about 2 to 3 stops down from the maximum, so in this case probably F5.6 or F8, although whether we can stop down to F8 without more degradation from other factors is another question. So at ISO 100 - and a desire for 1/60th - and F5.6 - you'll probably get a very under-exposed image ... so the only other alternative is to increase the ISO.

Using high ISO settings scares a lot of people - BUT - (and it's a BIG BUT) - it's usually NOT AN ISSUE if you remember 2 simple rules: (1) Don't under-expose the shot, because this will reveal MORE noise when you adjust the capture in post-processing, and (2) Frame the shot so that you don't need to crop excessively - this is because noise is small and to a large degree it averages out, so although it will be quite obvious if viewed at 100% magnification on a computer screen, it generally won't be visible (or at least obvious) in a real world print (unless you've printing exceptionally large prints).

I used the biggest aperture available to get as much sharpness as possible, yet the previous post also suggests to use lower apertures :/

Some high-quality lenses also perform well wide open, but aim for 2 to 3 stops down from maximum aperture for best sharpness. Also, whilst on the subject of sharpness, don't forget that correct image sharpening plays a MUCH bigger part in the final image than any inherant sharpness from the lens (assuming a correctly functioning lens though, which I don't think yours is).

I believe I used 'Evaluative Metering' and focused on the girl who is holding the baby. I believe she has a middle-grey-ish skin :P

Evaluative metering evaluates the entire scene, so doesn't matter what you focus on.

I am staying at my hometown where I don't have my desktop computer. If I had Canon's software I could tell you precisely the spots that were in focus.

With the depth of field you have, it wouldn't have made any difference.

Lens is a Tamron 28/80. I know it's far from being the best, but it was a bundle I saw on ebay that matched my budget :/

I'm afraid that Tamron doesn't have the best name in the industry for quality and performance. Enough said!

I'll get a few shots this coming weekend and check their sharpness on the sides. If it is indeed a lens performance issue, what lens would you suggest as a replacement?

It's going to depend on your budget. If you REALLY want to see quality from a lens, look at a Canon EF24-70 F:2.8L USM, but sit down before you look at the price. Thankfully there are good and cheaper alternatives, but as I use only L-Series, I'll let someone else chip in with some thoughts on these.

Edit: I ALWAYS have a UV filter. Could the filter be the cause?

Dave said "No", which I agree is most likely - however - it is technically possible if you've got a faulty one (I've seen a similar thing). Extremely unlikely, but easy to test.

[pwnage101]

Regarding Tamron 28-80 f3.5-5.6 Aspherical:

I'm afraid that Tamron doesn't have the best name in the industry for quality and performance. Enough said!

True, but plastic lensbabies perform better than the image in the OP.
See flickr.
See blank. (hehe, I intended to provide multiple links but nobody seems to use tamrons )

Regarding UV filter:

Dave said "No", which I agree is most likely - however - it is technically possible if you've got a faulty one (I've seen a similar thing). Extremely unlikely, but easy to test.

Absolutely true. But I would like to add that it could also be a faulty lens. Again, this is easy to test. Take off the filter and if it still looks blurry, then it's either a faulty lens or not the Tamron 28-80 f3.5-5.6 I searched on google.

[Colin Southern]

True, but plastic lensbabies perform better than the image in the OP.

In my opinion, that would be about the only thing a lensbaby would outperform, but I digress ...

[leansoli]

Hi Leandro,
Welcome to the CiC forums from me, good to have you join us, although ideally under better circumstances would have been preferable.

You may need to apply some noise reduction to iso 400 and above if viewing/printing large, especially if cropping down from the full size, but that is far, far easier and less damaging than this kind of effect.

Iso 400 definitely shouldn't "give you a chill"; I regularly shoot at 800 - 2000, or higher.

Is it possible you have mis-read "bigger" somewhere and applied it to the diameter of the hole instead of the f/ number?
Wide open on 95% of lenses will not produce the sharpest overall image, as many have said - it is a major 'feature' of lenses that the centre is sharper than the edges wide open. Whereas at f/16 or f/22, whatever the minimum diameter is, they are generally most evenly sharp, HOWEVER not as good as a mid-range stop like f/8, which usually balances good/better centre sharpness with acceptable edge sharpness.

No

Not really, what would be better than be here to learn? (probably be on the other end and 'teach', but have a long way until that) The lens performance is just that, a bad lens that may need a replacement

Indeed, when I speak of big apertures, I mean really low F stops. Is it technically wrong to speak about big apertures rather than speak about low f stops?

Another discussion for another day is that I wonder why this hasn't changed throughtout history, considering an "f stop" is a dimensionless number. I know f stops are here from day 0, but with the digital stuff reaching to loads of people, photography industry could have thought this out...

I noticed what you mention in previous posts: lens is better at min f stop + 1 or 2 stops. Will put this in practice and will not be affraid to raise ISO if can't balance with speed value

Hi Leandro,
No worries - it's what we do best here

Thanks for that. It's always nicer to be able to say something like "Hi John" rather than "Hi ForceToken6697" (except in this case your name is Leandro, so calling you John would just be silly!)

For landscape - shot from a tripod - this is almost mandatory. In situations where camera shake and/or subject motion may degrade an image then it's not so clear cut ...

Everything in photography is a compromise; too low a shutterspeed and you risk your image being degraded by camera shake or subject motion - too wide an aperture (low "F" number) and you risk your image being degraded by an insufficient depth of field or operating the lens in an area where it's performance is "less than ideal" - too high an ISO and you risk your image being degraded by noise and/or having issues with the reduced dynamic range (a topic for another day), so we need to work out what's the best combination.

In your case, you need to keep the shutterspeed high enough to freeze any inadvertant movement, and minimise any camera shake ... so personally I'd be aiming for at least 1/60th at this focal length. In terms of aperture, it's generally accepted that the "sweet spot" for most lenses is about 2 to 3 stops down from the maximum, so in this case probably F5.6 or F8, although whether we can stop down to F8 without more degradation from other factors is another question. So at ISO 100 - and a desire for 1/60th - and F5.6 - you'll probably get a very under-exposed image ... so the only other alternative is to increase the ISO.

Using high ISO settings scares a lot of people - BUT - (and it's a BIG BUT) - it's usually NOT AN ISSUE if you remember 2 simple rules: (1) Don't under-expose the shot, because this will reveal MORE noise when you adjust the capture in post-processing, and (2) Frame the shot so that you don't need to crop excessively - this is because noise is small and to a large degree it averages out, so although it will be quite obvious if viewed at 100% magnification on a computer screen, it generally won't be visible (or at least obvious) in a real world print (unless you've printing exceptionally large prints).

Some high-quality lenses also perform well wide open, but aim for 2 to 3 stops down from maximum aperture for best sharpness. Also, whilst on the subject of sharpness, don't forget that correct image sharpening plays a MUCH bigger part in the final image than any inherant sharpness from the lens (assuming a correctly functioning lens though, which I don't think yours is).

Evaluative metering evaluates the entire scene, so doesn't matter what you focus on.

With the depth of field you have, it wouldn't have made any difference.

I'm afraid that Tamron doesn't have the best name in the industry for quality and performance. Enough said!

It's going to depend on your budget. If you REALLY want to see quality from a lens, look at a Canon EF24-70 F:2.8L USM, but sit down before you look at the price. Thankfully there are good and cheaper alternatives, but as I use only L-Series, I'll let someone else chip in with some thoughts on these.

Dave said "No", which I agree is most likely - however - it is technically possible if you've got a faulty one (I've seen a similar thing). Extremely unlikely, but easy to test.

I was not aware of 'Simple rule °1'. I do keep in mind the other rule though, unless cropping is absolutely necesary (like somebody was begining to walk in front of you, etc).

I was also aware about Tamron's when I bought this kit, but I bought this thinking of a starter kit (consider I live in a country that currently has a 4 to 1 money conversion :/ )

I am fairly sure you all know these sites, but perhaps, some day, someone that reaches this thread will find them useful:

DPReview
Pixel Peeper

Regarding Tamron 28-80 f3.5-5.6 Aspherical:

True, but plastic lensbabies perform better than the image in the OP.
See flickr.
See blank. (hehe, I intended to provide multiple links but nobody seems to use tamrons )

Regarding UV filter:
Absolutely true. But I would like to add that it could also be a faulty lens. Again, this is easy to test. Take off the filter and if it still looks blurry, then it's either a faulty lens or not the Tamron 28-80 f3.5-5.6 I searched on google.

That's the lens, and I have sharp shots from last year. I'll go check all this this coming weekend.
Thanks for all this info!

[Katy Noelle]

This is such a helpful and interesting thread. Thank you, all, for taking the time to contribute to it.

[Dave Humphries]

Indeed, when I speak of big apertures, I mean really low F stops. Is it technically wrong to speak about big apertures rather than speak about low f stops?

No, I don't believe there's a right or wrong to it.

Just that we have to be careful to be completely unambiguous when writing about it in places like this, or people that are new to the whole lens thing can easily read something and come away with the opposite understanding to that which the author had intended.

Unfortunately, not everywhere on the world wide web takes as much care as we do and one person reading (thinking in f/numbers) can get mis-led by an author (thinking in diameters) writing "larger" or "bigger", or vice versa

Better to say "low/high" for f/number and "narrow/wide" for aperture - the latter meaning diameter

It's a blooming minefield

[arith]

Regarding Tamron 28-80 f3.5-5.6 Aspherical:

True, but plastic lensbabies perform better than the image in the OP.
See flickr.
See blank. (hehe, I intended to provide multiple links but nobody seems to use tamrons )

Regarding UV filter:


Absolutely true. But I would like to add that it could also be a faulty lens. Again, this is easy to test. Take off the filter and if it still looks blurry, then it's either a faulty lens or not the Tamron 28-80 f3.5-5.6 I searched on google.

I've got a Tamron, the Canon at twice the price is better, but the Tamron your talking about is cheap like the cheap tacky Canon 28-80 with plastic mount.

But if you've got an L type you don't have to do or know anything, just point and shoot and it will turn out alright. I have to know about resolution, the resolution or Nyquist of my camera 2250 LW/PH, and that I can do just as well as the chap using program mode with an L type.

I think it will be very nice to have an L type, but I can't afford one and not all Canon lenses are good.

Obviously the lens here is faulty, or extremely bad at edges, so stop it down a bit.

[benm]

I will have to agree with those who say the lens may be at fault. I also noticed that the stone columns in the photo have severe distortion, maybe from having tilted the camera a little. If you test the lens you can determine what its problems are and how to minimize them. With your lens I might try using a longer focal length and backing up appropriately (if you have the room).