| Author | Thread |
|
|
09/10/2007 02:24:28 PM · #1 |
Hello. I am sorry for maybe bringing out the old stinky-rotten horse, only to beat at it again. But I am wanting to be a student of sharp, crisp images, only to be disappointed shot after shot.
I look at the top images (on this site) for the lens I have that is said to be sharp. Canon 50mm f/1.8 II . These images are too tack-on and this is the effect I want to obtain.
See:
&:
I want to know general rules in obtaining such images. Is it all lighting? F-stops? (Is there really a big difference in sharpness with different f-stops?) How do I create an ideal situation around my subjects to allow for my gear to capture the clarity our eyes bedazzle us with daily with little notice?
On my quest for clarity and sharpness, I plan to move onto a better lens. I am thinking about the Canon 17-40mm f/4L. It seems like it can be had for around $600 bucks. It will take some months to save for. I think it would be a great "walk around" lens to boot, would anyone disagree?
Now, how can I practically-practice for this big wig of a lens with the one I currently have. (Canon 50mm f/1.8 II)
Thanks for your time! |
|
|
|
09/10/2007 02:32:52 PM · #2 |
Originally posted by jfriesen:
I want to know general rules in obtaining such images. Is it all lighting? F-stops? (Is there really a big difference in sharpness with different f-stops?) How do I create an ideal situation around my subjects to allow for my gear to capture the clarity our eyes bedazzle us with daily with little notice? |
Good lighting, steady camera (tripod if needed), faster shutter speeds...
Any examples of what you thought should have been sharp and wasn't (from your portfolio, etc...)? |
|
|
|
09/10/2007 02:33:00 PM · #3 |
Smart sharpen at around 50% strength, .03 radius, remove gaussian blur. Do it after your final resize for dpc. Sometimes, do it twice. That will crisp it up perfectly (you can play with the strength settings but is usually all it takes).
Stop your lens down a bit to get a sharper image as well, but that won't make much of a difference noticeable at 640px. |
|
|
|
09/10/2007 02:44:50 PM · #4 |
Originally posted by glad2badad:
Any examples of what you thought should have been sharp and wasn't (from your portfolio, etc...)? |
I don't see anything in my profile that is interestingly sharp.
Originally posted by routerguy666:
Smart sharpen at around 50% strength, .03 radius, remove gaussian blur. Do it after your final resize for dpc. Sometimes, do it twice. That will crisp it up perfectly (you can play with the strength settings but is usually all it takes).
Stop your lens down a bit to get a sharper image as well, but that won't make much of a difference noticeable at 640px. |
Thanks for the tip.. I will try this! |
|
|
|
09/10/2007 02:56:19 PM · #5 |
Notice both of those shots were done at faster than 1/100 shutter speed and ISO 100 with the lens stopped down. Pretty much any lens will be sharp if you have good enough light for those settings.
If you don't have lights push ISO as far up as possible to get at least (1/focal length of lens) shutter speeds (ie 1/80) for the 50mm plus 1.5 crop factor and at least 1 stop above wide open (best if you can go all the way up to your lens' sweet spot). I won't hesitate to go up to ISO 1600 rather than letting my shutter speed drop to low.
Light is the most important aspect so read  Idnic's Portrait Lighting -- Learning Thread |
|
|
|
09/10/2007 11:51:38 PM · #6 |
| Thanks for the tips guys! |
|
|
|
09/10/2007 11:55:16 PM · #7 |
Originally posted by routerguy666:
Smart sharpen at around 50% strength, .03 radius, remove gaussian blur. Do it after your final resize for dpc. Sometimes, do it twice. That will crisp it up perfectly (you can play with the strength settings but is usually all it takes).
|
Or higher. A lot of my recent entries are at about 115% at .5 pixels.
|
|
|
|
09/11/2007 12:17:01 AM · #8 |
Originally posted by glad2badad:
Good lighting, steady camera (tripod if needed), faster shutter speeds...
|
and appropriate aperture. Most lenses will be sharpest a couple of stops away from wide open. f5.6, or f8 on most lenses would close to the sweet spot. Wide open or all the way shut down aren't usually as good.
The things you mentioned certainly matter a lot too. Tripod being the first, most important element, probably. When I've had an actually sharp image, it usually doesn't need much, if any sharpening before submitting here.
Message edited by author 2007-09-11 00:17:45.
|
|
|
|
09/11/2007 01:27:15 AM · #9 |
| If you are using the 50/1.8 and want sharp, do not consider using f/1.8 or f/2.2. They are quite soft on that lens. After that, at least on the copy I have, it starts to get sharper. The sharpest, naturally, is f/8, but you don't always get the photographic intent you may want at that aperture. |
|
|
|
09/11/2007 01:38:47 AM · #10 |
In regards to shutter speed the rule of thumb I always use for a stationary subject is:
Shutter Speed = 1/Focal Length
So if my focal length is 180mm then I want to have a shutter speed at least 1/180.
|
|
|
|
09/11/2007 08:58:27 AM · #11 |
On the 50mm 1.8 I've found the sharpest F setting to be around 4.5-5.6. I think this was confirmed on Photozone but I can't be sure so don't quote me, I'll just give it a loose endorsement.
And the 17-40mm? Well, I borroed my dads for a while and I wasn't blown away. It's f4 which i didn't like too much and it honestly wasn't the sharpest lens I've used in that focal length, and for the price I think the Tamron I bought after is a better bet- cheaper, faster, not as strong structurally, but to me seems sharper at f4. However, it does not have the red ring around it, but it wont let you down. But its your money, your decision. |
|
|
|
09/11/2007 10:17:07 AM · #12 |
I concur on the Tamron 17-50 and on everything everyone else has said about smaller than max. aperture, tripod, and shutter speed. As for "neat," try different settings on your camera. Canons tend to have less noise but are also less sharp, Nikons tend to go the other way, and I don't know about other brands. But those differences appear primarily in extreme circumstances.
|
|
|
|
09/11/2007 10:37:21 AM · #13 |
Well hey, I am thinking about getting the Tamron 28-75 instead, (I've heard GREAT things about it in comparison to the 17-40L) but that might not be wide enough for me.
Is the 17-50 better than the 28-75 to boot?
Edit: Or the 17-35 for that matter.
Message edited by author 2007-09-11 10:47:02. |
|
|
|
09/11/2007 10:53:31 AM · #14 |
They're pretty much the same lens but with different focal lengths. I got the 17-50 due to the wider angle, and i have another lens that gives me the telephoto end so that isn't an issue.
Dont forget that with the 1.6x crop factor the 24mm may not be wide enough for what you're after.
My vote is with the 17-50mm. |
|
|
|
09/11/2007 10:56:16 AM · #15 |
Originally posted by Tez:
They're pretty much the same lens but with different focal lengths. I got the 17-50 due to the wider angle, and i have another lens that gives me the telephoto end so that isn't an issue.
Dont forget that with the 1.6x crop factor the 24mm may not be wide enough for what you're after.
My vote is with the 17-50mm. |
That's what I am thinking too..
But, I just noticed this was taken with the 17-40L
If I was out with *my* 17-50, could I get the same results with color replication and clarity?
|
|
|
|
09/11/2007 11:15:37 AM · #16 |
Certainly.
Its not as if that image was lifted straight from the camera like it is. Check out Photozone.de for some reviews of the 17-50 Tamron and look at the charts on resolution, you'll be pleasantly suprised. The only lens I can name right now that's sharper and in the same focal length is the 18-55 IS USM and it's about $1200 or so!
Now if you had the same lighting, subject, location, camera settings and timing as the shot you mentioned then I'm fairly certain they would be 99% identical in terms of color and clarity.
If your heart is set on the 17-40mm then by all means buy it, its great. But in my eyes not the best. Of course the best way is to go to a camera store and say "look, i have some money and i'm thinking of getting the 17-50 2.8 tamron, help me out". See what you think. I wouldnt mention youre thinking of getting either/or because they'll push the most expensive option.
When I was wondering about a walkaround lens i looked EVERYWHERE- Fred Miranda, photozone, slrgear.com, here and magazines because after all, it's a lot of money so youd better get it right. I have full confidence in the Tamron and would recommend it to anyone. Likewise if mine was destroyed, I would buy another one immediately.
|
|
|
|
09/11/2007 11:36:19 AM · #17 |
Does the 20D have an APC sized sensor? If so, then for hand holding, the 1/focal length = good shutter speed would need adjusting to 1/1.5 focal length because of the effective difference in focal length. Just something to think about.
Sharpen in PP, and then again as the last step before uploading after sizing to required size. It makes a huge difference.
It also helps a lot to resize in small increments, 500 or so pix in width instead of a one jump resize to finished size. This may be what is causing the problem you are seeing. It makes a lot of difference. I was surprised the first time I did it that way.  Kirbic told me about doing that, and it works.
|
|
|
|
09/11/2007 12:28:41 PM · #18 |
Originally posted by jfriesen:
But, I just noticed this was taken with the 17-40L
If I was out with *my* 17-50, could I get the same results with color replication and clarity? |
Here are reviews of the two on slrgear.com. The Tamron gets a slightly higher rating for image quality. To me, that just means they're both excellent lenses.
Canon 17-40 4L
Tamron 17-50 2.8SP
So. For 2/3 the cost, you get an extra stop of light and an extra 10mm of zoom in the Tamron. FWIW, "SP" means Tamron's higher-end lenses, the way "L" does for Canon.
Originally posted by MelonMusketeer:
Does the 20D have an APC sized sensor? If so, then for hand holding, the 1/focal length = good shutter speed would need adjusting to 1/1.5 focal length because of the effective difference in focal length. |
I'm not sure that's right. You're not really magnifying 1.5 times as much. Rather, you're just not seeing the full image that the lens produces. So any jumpiness will be the same as a full-frame camera sees. Or so it seems to me, anyway.
Originally posted by MelonMusketeer:
Sharpen in PP, and then again as the last step before uploading after sizing to required size. It makes a huge difference.
It also helps a lot to resize in small increments, 500 or so pix in width instead of a one jump resize to finished size. This may be what is causing the problem you are seeing. It makes a lot of difference. I was surprised the first time I did it that way. Kirbic told me about doing that, and it works. |
I agree about when to sharpen, but I've read numerous times that at least from Photoshop CS2 on, the default resize works just fine, without needing to do it in steps.
Message edited by author 2007-09-11 12:36:41.
|
|
|
|
09/11/2007 12:45:24 PM · #19 |
Originally posted by jfriesen:
But, I just noticed this was taken with the 17-40L
If I was out with *my* 17-50, could I get the same results with color replication and clarity? |
I think it's silly to make lens comparisons based on little tiny images like this to be honest, unless that is going to be the sort of thing you will end up doing with all your shots. |
|
|
|
09/11/2007 01:01:49 PM · #20 |
Originally posted by levyj413:
Originally posted by MelonMusketeer:
Does the 20D have an APC sized sensor? If so, then for hand holding, the 1/focal length = good shutter speed would need adjusting to 1/1.5 focal length because of the effective difference in focal length. |
I'm not sure that's right. You're not really magnifying 1.5 times as much. Rather, you're just not seeing the full image that the lens produces. So any jumpiness will be the same as a full-frame camera sees. Or so it seems to me, anyway.
|
Hmm, I think melon musketeer is right. Yes the physical jumpiness distance is the same, but for a smaller sensor it will be a larger distance on the image compared to a full sensor. Thats a terrible explanation.
How about, hypothetically speaking:
Say for a full frame sensor, moving the camera 1mm or less during the exposure gives acceptable focus. That is, on the final image, the slight blur caused by that 1mm movement is indistinguishable to the human eye compared to the full resolution image.
On a 1.5x crop sensor, that 1mm deviation will cause 1.5 times as much motion blur on the image. So when looking at the full resolution image the blur will not be visible.
Apologies if that is rubbish or incomprehensible. Anyone got a better explanation? |
|
|
|
09/11/2007 01:02:50 PM · #21 |
Originally posted by levyj413:
I'm not sure that's right. You're not really magnifying 1.5 times as much. Rather, you're just not seeing the full image that the lens produces. So any jumpiness will be the same as a full-frame camera sees. Or so it seems to me, anyway.
|
Don't forget that, although you are using just a portion of a photograph as you said, you will then enlarge that portion to view it at the same size as one would any other photo, so any camera-shake blur will be magnified, too. I seem to remember seeing this all equated to FOV, rather than focal length, but can't remember where.
As a bit of an example, consider a fictional camera with a small enough sensor where a 1mm lens, owing to the 'crop factor' gives an 'effective' 35mm equivalent of 200mm. One would not be able to take a 1 second exposure hand held.
All that aside, this is a rule of thumb. Everyone's mileage will vary.
Edit: Beaten to it again.
Message edited by author 2007-09-11 13:03:33.
|
|
|
|
09/11/2007 02:07:32 PM · #22 |
Originally posted by Mr_Pants:
Don't forget that, although you are using just a portion of a photograph as you said, you will then enlarge that portion to view it at the same size as one would any other photo |
Thanks to both you and latentflip for engaging on this. :)
I think you nailed it - it depends on how many pixels you're talking about, not the physical size of the sensor.
I think a camera where the image is 2000 pixels across will show only half the movement of one 4000 pixels across, regardless of whether there's a crop factor or not. That is, the same physical distance across the sensor will cross half as many pixels. I think today's range of cameras has similar variation in how many pixels there are, certainly there are cameras with more than 1.5x the pixels in one dimension than others, which would more than make up for the crop factor.
So I think it depends on the camera.
But anyway, wasn't that old reciprocal rule of thumb based on 35 mm film? I don't know how well that translates to digital sensors anyway. What was the equivalent number of pixels in a 35mm frame? I guess that's the length of the frame divided by the grain size?
Ah, isn't it nice to find an academic discussion vs. arguing about defining the latest challenge? :)
|
|
|
|
09/11/2007 03:39:57 PM · #23 |
Originally posted by levyj413:
Originally posted by Mr_Pants:
Don't forget that, although you are using just a portion of a photograph as you said, you will then enlarge that portion to view it at the same size as one would any other photo |
Thanks to both you and latentflip for engaging on this. :)
I think you nailed it - it depends on how many pixels you're talking about, not the physical size of the sensor.
I think a camera where the image is 2000 pixels across will show only half the movement of one 4000 pixels across, regardless of whether there's a crop factor or not. That is, the same physical distance across the sensor will cross half as many pixels. I think today's range of cameras has similar variation in how many pixels there are, certainly there are cameras with more than 1.5x the pixels in one dimension than others, which would more than make up for the crop factor.
So I think it depends on the camera.
But anyway, wasn't that old reciprocal rule of thumb based on 35 mm film? I don't know how well that translates to digital sensors anyway. What was the equivalent number of pixels in a 35mm frame? I guess that's the length of the frame divided by the grain size?
Ah, isn't it nice to find an academic discussion vs. arguing about defining the latest challenge? :) |
Camera shake is a problem with regards to angular movement; if your movement covers 1 degree and your lens covers 100 degrees, that's 1% and it's fine. If your lens coverage is 33 degrees, the same amount of movement is 3% and that's not so good.
Crop factor is therefore highly relevant; Even in the film days, if you were using a 100mm lens and thought 1/100 was ok, sometimes when you cropped in on the image you'd find it wasn't okay after all, because you now had cropped to the equivalent of a 200 mm lens and this magnified and shake that was present.
So yes, on APS-C size sensors, the formula should be 1/1.5x the focal length to follow the same rule of thumb, and furthermore any time you are shooting hand held and plan to crop the image significantly, you need to ramp up shutter speed to compensate.
Or use a tripod of course...
R.
|
|
|
|
09/11/2007 05:51:25 PM · #24 |
| Good job bear_music, simplified what I was trying to say massively. Sometimes getting the contents of my brain into text is much harder than answering a problem in my brain :). |
|
|
|
09/11/2007 10:01:31 PM · #25 |
"Rule of thumb" certainly applies to me, as I sometimes think that I am all thumbs.
Good explanation Bear_Music
|
|
