125 replies to this topic

[whitey]

But I can not confirm that FF wideangles deliver poorer corners in general. I think we have been spoilt by FF lenses on cropped sensors where corners .

Julian

<{POST_SNAPBACK}>

Do you mean topside or through a dome port, Julian? Pretty much all lenses, and all the practical underwater options, will be a bit soft in the corners on full frame when shot wide open in above water use. If you want really sharp corners on the 5D, you're pretty much talking Zeiss 21, Zuiko 21 and Leica Elmarit 19mm - but I don't think you want to take any of these diving. In reality, the corner performance of something like the 17-40L would be fine for underwater use from the point of view of the lenses' optics.

I do wonder whether the softish corners (from the dome port, not the lens) that I see when shooting underwater on the 10D would be significnatly worse if shooting full frame. Even if Ikelite for example decides to house the 5D, I'm figuring that there's no experience available in using a port system like this for full frame. I guess I'm thinking that all the issues like nodal point etc would be magnified if we're using the full lens and not just the centre crop.

[Phil Rudin]

I suspect that the 1:1 claim of the Canon 60mm lens is based on the 35mm format, even though it mounts only to EF-S camera, thus yielding greater than 1:1.  Otherwise the degree of magnification would be less than the 1:1 of the 100 macro (which IS based on 35mm), and offer even further confusion to the lens lineup!

<{POST_SNAPBACK}>

If this is the case then Canon should step up and call the lens what it is a 1:1.6 macro.

Phil

[Jolly]

Rob,

as said, when leaving dome effects out for the first.
So what EF-S lenses would you suggest to deliver better corners than 17-40L, 20 f2.8, ... on full frame? Do you expect the 10-22 to be better on a high resolution 1.6 sensor (which doesn't exist from Canon) than the others on full frame?

As said, the dome is a factor as it requires DOF for wideangle corners. FF lenses call for large domes.

Julian

[Phil Rudin]

Rob,

as said, when leaving dome effects out for the first.
So what EF-S lenses would you suggest to deliver better corners than 17-40L, 20 f2.8, ... on full frame? Do you expect the 10-22 to be better on a high resolution 1.6 sensor (which doesn't exist from Canon) than the others on full frame?

As said, the dome is a factor as it requires DOF for wideangle corners. FF lenses call for large domes.

Julian

<{POST_SNAPBACK}>

Like the issue of life size (i.e. 1:1) full frame has nothing to do with 35 mm. Any lens that was designed for the sensor being used makes the sensor a full frame sensor. A 645 lens used with a 645 sensor is full frame. Canon has four EF-S lenses, 60 mm macro, 10-22, 17-85, 18-55. These lenses are designes for the so called 1.6 sensor and don't work on the 1.3 sensor or the 35mm sensor. That makes the 1.6 sensor full frame for those lenses. Nikon has six DX lenses that don't work on 35 mm, these lenses make the 1.5 sensor full frame. Olympus has nine digital only lenses and four more on the way. They only work on the 4/3 sensor and no 35 mm lens is even made that will fit the lens mount. That makes the Oly lens sensor combo clearly full frame. I admit that most camera users come from a 35 mm backgroung but that in no way makes 35 mm the standard by which all other formates should be judged. I have a Pentax 6 x7 camera for some types of work where I need a large neg. If I used that as the standared 35 mm would have a X2 crop factor. Regarding 35 mm sensor v 1.6 or 1.5 with like angle of view lenses. Go to Steve Frinks review of the 17 to 40L on a 35 mm sensor and the review of the Nikon 12 to 24mm
on the Nikon high end 12+ MP sensor. Results show the smaller sensor giving the better image.
I beleave this is because the 17 to 40 was designed for a film camera and the 12 to 24 was designed for the 1.5 chip. If you read the date regarding many of the high end designed for digital lenses you will find that the elements are of a diffrent design from film lenses. Also that the wider the film lens the worst the preformance of the lens in the corners, land or sea. See Steve Frink test of 16 to 35L mm v 17 to 40L on 35 mm full frame.

Phil

[Jolly]

The term full frame is used more than regularly for describing a DSLR recording surface of equal size to 35mm film, from the industry too. Of course you may argue if you find this term misleading.

The problem with larger image recording size for underwater is the shallow DOF because we shoot thru simple curved glass which actually is not a consistent optical underwater system. Again, the problem remains especially with wideangles in the corners as the dome’s virtual image is curved, not flat. Corners are closer in terms of focus distance. So they have to be covered by DOF. That’s why dome glass didn’t work well with medium format wideangles and things like the Ivanoff underwater optical corrector have been used instead. I agree that on an identical dome setup the APS-C lens (same FOV,etc.) performs better. But the reason is not the general opical disadvantage of the 35mm format, it's the hosuing optics which are very poor compared to any decent wideangle lens.
By the way, this does not apply to fisheyes because they focus in a manner the dome’s virtual image is shaped. Fisheyes have always been less critical underwater in the film past too.

What I was trying to say to Rob: A lot of people complain about 35mm sized sensors delivering poor (wideangle) corner performance. I think this is often meant in comparison to APS-C DSLRs with 35mm lenses because the weak corners are cut away. But you loose the FOV too. IMO you experience the same with an APS-C image chip and APS-C lenses as they capture the entire lens’s image circle too, including critical corners at the edge of the FOV the lens was built for (we talk about extreme wideagnles around 100° FOV). I’ve seen poor corners with both formats, 35mm with 35mm lenses and APS-C with APS-C lenses. And I am not sure that every APS-C zoom must outperform any 35mm lens (prime/zoom) in general.

Don't know the tests you have mentioned. I guess they are in conjunction with underwater usage incl. domeports? or lens only tests? Maybe you could post a link? I found this quiet interesting:
http://www.wlcastlem...17-40/union.htm

Julian

edit/ps: 17-40 for designed for digital 35mm sensors. The older 16-35mm was film only in mind.

[Phil Rudin]

The term full frame is used more than regularly for describing a DSLR recording surface of equal size to 35mm film, from the industry too. Of course you may argue if you find this term misleading.

The problem with larger image recording size for underwater is the shallow DOF because we shoot thru simple curved glass which actually is not a consistent optical underwater system. Again, the problem remains especially with wideangles in the corners as the dome’s virtual image is curved, not flat. Corners are closer in terms of focus distance. So they have to be covered by DOF. That’s why dome glass didn’t work well with medium format wideangles and things like the Ivanoff underwater optical corrector have been used instead. I agree that on an identical dome setup the APS-C lens (same FOV,etc.) performs better. But the reason is not the general opical disadvantage of the 35mm format, it's the hosuing optics which are very poor compared to any decent wideangle lens.
By the way, this does not apply to fisheyes because they focus in a manner the dome’s virtual image is shaped. Fisheyes have always been less critical underwater in the film past too.

What I was trying to say to Rob: A lot of people complain about 35mm sized sensors delivering poor (wideangle) corner performance. I think this is often meant in comparison to APS-C DSLRs with 35mm lenses because the weak corners are cut away. But you loose the FOV too. IMO you experience the same with an APS-C image chip and APS-C lenses as they capture the entire lens’s image circle too, including critical corners at the edge of the FOV the lens was built for (we talk about extreme wideagnles around 100° FOV). I’ve seen poor corners with both formats, 35mm with 35mm lenses and APS-C with APS-C lenses. And I am not sure that every APS-C zoom must outperform any 35mm lens (prime/zoom) in general.

Don't know the tests you have mentioned. I guess they are in conjunction with underwater usage incl. domeports? or lens only tests? Maybe you could post a link? I found this quiet interesting:
http://www.wlcastlem...17-40/union.htm

Julian

edit/ps: 17-40 for designed for digital 35mm sensors. The older 16-35mm was film only in mind.

<{POST_SNAPBACK}>

Regarding the industry use of the term full frame the industry also uses the term life size or 1:1 in regard to 35 mm as well, I have no argument regarding this other than to say it is in fact misleading.
It was not my intent to suggest that all digital specific lenses are better than all film lenses.
I just wanted to point out that the evidence is becoming stronger that lenses designed for a digital sensor type appear to be gain ground over lenses designed for film. Which leads to the question will 35 mm film lenses need to be replaced by 35 mm digital lenses for the 35 mm chip to achieve max image quality?
The Frink tests can be found a seacamusa.com in the Tech Notes section.
Regarding corner sharpness, all things being equal should I expect that when my Olympus 7 to 14 mm zoom (14 to 28 mm 35 equiv.) dome port arrives in the next month or so that I should expect it to be the sharpest lens of all in the corners because it has the smallest SLR sensor and therefore the greatest DOF?

[Jolly]

Yes, I agree there is some work to do with both formats. 35mm wideangles should be overworked to meet nowadays’ digital 35mm SLR bodies (5D / 1Ds) requirements. An increased choice on APS-C lenses (more zooms, fast primes, etc.) is important too.

As a Canon shooter, I would prefer the 35mm lens choice over EF-S - from what is available now. As said, I am really afraid EF-S won’t become an extensive lens line-up like Nikon’s DX series.

4/3” 7-14mm: Well, I don’t expect Olympus will ship you an adapted Seacam Superdome with E-300 housing mount
But yes, if everything else would be absolutely identical, 4/3” wideangles have a dome corner advantage in terms of DOF.

Julian

[Rocha]

Did you guys check the 5D samples at the Canon site? The colors are awesome, but I noticed soft corners on the landscape shot. I don't know which lens was used, the exif only says 17mm, I am guessing the 17-40.

http://web.canon.jp/...d_sample-e.html

[Drew]

Did you guys check the 5D samples at the Canon site? The colors are awesome, but I noticed soft corners on the landscape shot. I don't know which lens was used, the exif only says 17mm, I am guessing the 17-40.

http://web.canon.jp/...d_sample-e.html

<{POST_SNAPBACK}>

EXIF says 17-40.
The grass looks like mushy water color in the corners. A little better nearer the center. Actually that landscape shot is not that great... but the portraits are amazing. Pretty typical of digital outresolving WA lenses.

[Jolly]

this sample doesn't look good at all. I really hope this is not representive for the 5D + 17-40mm. I am not surprised by the corner softness, it's 17mm (93°) and somehow all extreme wideangles produces such corners, especially on this kind of high resolution cameras. But the rendered colors and details look strange. I sometimes wonder why manufactures put such samples online to promote their gear, even if they are real . Most worse I have seen is a Olympus 7-14mm sample from the manufacture itself:
http://www.olympus-e...ge/sample08.jpg

... and this is just a 5 MP camera not demanding this 1800 bucks lens to strong. Like the Olympus lens, I hope the 5D/17-40 will do better in real life too. Otherwise it's a bit dissapointing.

Julian

[Phil Rudin]

Yes, I agree there is some work to do with both formats. 35mm wideangles should be overworked to meet nowadays’ digital 35mm SLR bodies (5D / 1Ds) requirements. An increased choice on APS-C lenses (more zooms, fast primes, etc.) is important too.

As a Canon shooter, I would prefer the 35mm lens choice over EF-S - from what is available now. As said, I am really afraid EF-S won’t become an extensive lens line-up like Nikon’s DX series.

4/3” 7-14mm: Well, I don’t expect Olympus will ship you an adapted Seacam Superdome with E-300 housing mount 
But yes, if everything else would be absolutely identical, 4/3” wideangles have a dome corner advantage in terms of DOF.

Julian

<{POST_SNAPBACK}>

The Seacam dome would be beyond great and you could throw in a 180 enlarged finder at the same time. My guess is that the dome will be a 170 mm Athena glass on the Oly mount with the removable extension tube for the zoom lens. The FF 8 mm fisheye coming in Jan 06 is a 180 lens and projected at around 153 degrees under the dome without the tube and focus to 2 mm from the dome.
For a $1000.00 housing the viewfinder is not half bad.
you can see my post regarding this housing in the E-300 post in this forum.

Phil

[BradDB]

Herb - 1:1 traditionally means that the image on the 35mm fram is the same size as the subject, so the APS crop is actually larger than 1:1.  This leaves the question whether the EF-S 60mm lens is 1:1 on the EF-S sensor, or really greater than 1:1. 

Ok, maybe I can help resolve this issue.
I have a Canon 350D with a Canon 60mm EF-S lens sitting next to my desk that I use for work.
I set the camera for f8, 1/100, iso1600. I set the lens to manual, minimum focus (all the way left). I then photographed this ruler at what I estimated to be the sharpest focus I could achieve by moving the camera in and out. Here is the result:



This shot has not been cropped. I've only resized it to fit this forum.

Units are in inches. I estimate the total width to be 13.5 x 1/16 inches = 0.844 inches = 21.4 mm
The 1.6 crop sensor in the 350D is 22.2 mm across but not all of it is used.

I think this answers the question concerning 1:1 and the relevance of 35 mm equivalency. A 35 mm equivalency is not used when rating this lens. The image is the same size as the cropped digital sensor at 1:1.

-Brad

[Kasey]

Ok, maybe I can help resolve this issue.
I have a Canon 350D with a Canon 60mm EF-S lens sitting next to my desk that I use for work.
I set the camera for f8, 1/100, iso1600. I set the lens to manual, minimum focus (all the way left). I then photographed this ruler at what I estimated to be the sharpest focus I could achieve by moving the camera in and out. Here is the result:



This shot has not been cropped. I've only resized it to fit this forum.

Units are in inches. I estimate the total width to be 13.5 x 1/16 inches = 0.844 inches = 21.4 mm
The 1.6 crop sensor in the 350D is 22.2 mm across but not all of it is used.

I think this answers the question concerning 1:1 and the relevance of 35 mm equivalency. A 35 mm equivalency is not used when rating this lens. The image is the same size as the cropped digital sensor at 1:1.

-Brad

<{POST_SNAPBACK}>

Thanks Brad! I'm surprised that canon departed from the 35mm convention! So the 60mm lens gives signficantly less magnification than its FF cousin.

[herbko]

Thanks Brad!  I'm surprised that canon departed from the 35mm convention!  So the 60mm lens gives signficantly less magnification than its FF cousin.

<{POST_SNAPBACK}>

Kasey, I don't understand what you mean by this. Canon did what I think is the correct way to state a spec: 1:1 magnification means the image is the same 22mm as the subject. The 60mm lens has exactly the same magnification as the 1:1 100mm macro which was made for 35mm cameras.

A subject 22mm completely fills the 22mm frame rather than a 35mm subject. If he makes a 8x12 print from this shot the 22mm subject will come out 12 inches high, in the 35mm case a 35mm subject will become 12 inches high. The magnification would be GREATER if Canon were to state this in 35mm terms. See why it's so confusing to keep translating things to 35mm frames.

[Kasey]

Herb - if a 35mm lens is rated at 1:1 (as is the 100mm), it will provide greater than 1:1 magnification when mounted to APS - agreed? On APS the image circle is cropped leading to an apparent higher magnification.

All my micro Nikkors are rated at 1: 1 (35mm) but yield greater magnification on the cropped sensor.

In turn, the EF-S 1:1 lens offers only 1:1 magnification on the cropped chip. It provides less magnification than mounting a 35mm macro lens to the same body.

I think that this is confusing as a consumer - mounting the 1:1 100mm lens to my 20D yields greater magnification than mounting the 1:1 60mm EF-s to the same camera.

[herbko]

Herb - if a 35mm lens is rated at 1:1 (as is the 100mm), it will provide greater than 1:1 magnification when mounted to APS - agreed?  On APS the image circle is cropped leading to an apparent higher magnification. 

All my micro Nikkors are rated at 1: 1 (35mm) but yield greater magnification on the cropped sensor.

In turn, the EF-S 1:1 lens offers only 1:1 magnification on the cropped chip.  It provides less magnification than mounting a 35mm macro lens to the same body.

I think that this is confusing as a consumer - mounting the 1:1 100mm lens to my 20D yields greater magnification than mounting the 1:1 60mm EF-s to the same camera.

<{POST_SNAPBACK}>

Kasey. I think I see your point of confusion. Your last statement is incorrect. You will get exactly the same image that Brad got (no more no less magnification) if you mount your 100mm lens on your 20D. I've done the test with the same 100mm lens on my 300D. The 60mm and the 100mm has the same max magnification. Just ignore the fact that it's an EF-s lens and not usable on a FF camera.

Canon, the lens maker, is giving the correct spec 1:1 on the two lenses which has identical max magnification. The size of the "usable image" is not part of the spec; it's implicitly know that it's 35mm in one case and 22mm in the other.

The confusion comes from the fact that you're interested in magnification in the final print, which involves magnifying the image on the sensor to print size. You are making a larger magnification to go from the 22mm sensor to print vs the 35mm sensor. This is not included in the spec of the lens. The lens magnification spec only covers the part of going from your subject to the sensor (that's the only function of the lens); magnification from sensor to print is NOT included.

[Rocha]

Herb - if a 35mm lens is rated at 1:1 (as is the 100mm), it will provide greater than 1:1 magnification when mounted to APS - agreed?  On APS the image circle is cropped leading to an apparent higher magnification. 

All my micro Nikkors are rated at 1: 1 (35mm) but yield greater magnification on the cropped sensor.

<{POST_SNAPBACK}>

No, when you crop (be it in a sensor or in photoshop) you don't increase magnification, you change the field of vision. So, 1:1 is 1:1 in 35mm and 1:1 in 20mm or whatever the measure of your sensor is, we just change the field of view.

In other words, if you take the same lens (say a non EFS 60mm), and make the ruler test (the same already posted here), you will photograph a 22mm (or 21.4 as demonstrated) long stretch of the ruler with the 1.6 cropped camera (dRebel or 20D), meaning that it is a 1:1 lens because it photographs an object that has the same size of the sensor. If you take the exact same lens and mount it on a full frame (35mm sensor) camera you will photograph a 35mm long stretch of the same ruler, still meaning that the lens is 1:1. You didn't change the magnification, just the field of view.

EDIT: Herb, you posted when I was typing, beat me! We are giving the same answer here.

[whitey]

The term full frame is used more than regularly for describing a DSLR recording surface of equal size to 35mm film, from the industry too. Of course you may argue if you find this term misleading.

What I was trying to say to Rob: A lot of people complain about 35mm sized sensors delivering poor (wideangle) corner performance. I think this is often meant in comparison to APS-C DSLRs with 35mm lenses because the weak corners are cut away. But you loose the FOV too. IMO you experience the same with an APS-C image chip and APS-C lenses as they capture the entire lens’s image circle too, including critical corners at the edge of the FOV the lens was built for (we talk about extreme wideagnles around 100° FOV). I’ve seen poor corners with both formats, 35mm with 35mm lenses and APS-C with APS-C lenses. And I am not sure that every APS-C zoom must outperform any 35mm lens (prime/zoom) in general.

Don't know the tests you have mentioned. I guess they are in conjunction with underwater usage incl. domeports? or lens only tests? Maybe you could post a link? I found this quiet interesting:
http://www.wlcastlem...17-40/union.htm

Julian

edit/ps: 17-40 for designed for digital 35mm sensors. The older 16-35mm was film only in mind.

<{POST_SNAPBACK}>

No, Julian, he may not argue if he finds the term misleading. It's bloody well called full frame by everyone, so he'd better get used to it! FWIW, tropical, your pentax 6*7 is crop sensor also, given that I shoot MF at 6*9.

Anyway, I've previously posted brick wall shots on Wetpixel of the 17-40L at full frame re: corner sharpness. Please don't make me do it again! Wetpixel briefly consisted of nothing but peoples' wall shots - personally I like looking at fish and reefs and stuff better.

The point is this. A lens like the 17-40L on a full frame DSLR will peform very nicely, if you stop it down a little. If you pixel peep, the corners are still soft at f8 to f11, but not so you'd notice on an underwater shot (compared to the distortion and softness once a port is involved). It's not a problem of digital lens vs film lens - it's a problem of the sensors on the Canon 1Ds series (and now hopefully the 5D) producing files of medium format quality, through a lens designed for 35mm film use. In short, Canon's state of the art full frame sensors outressolve nearly all lenses. The three I mentioned above are some of the few (third party) lenses that are good enough to match the current generation of sensors. (They're not really suitable underwater as they're all manual focus/manual aperture - and I don't think the UW pics would be any sharper anyway).

I use the 17-40L on the 1Ds for 80% of my topside shooting. It's not perfect (and I'm looking for a sharper wide lens as we speak), but it's pretty damn good IMHO.

[Paul Kay]

Just to wade in with my own OBSERVATIONS (not derived from anything other than using the lenses). I've used both the Nikon 12~24 and Canon 17~40, topsides and below. Out of the two lenses I prefer the 17~40 for a variety of reasons, although both exhibit corner softness at times depending on a whole host of factors (it really is not as simple as what size sensor the lens is used on as there are cosiderations regarding sensor pixel density, pixel size, etc., etc.! - The theory is not over complex but there are a lot of variables).

But to get back to the REAL WORLD. My observations on wide-angles in general is that the high MPixel cameras will show flaws in virtually all of them whether used underwater or not. I currently use a Canon EOS1DS - not the MkII - and have consciously decided NOT to upgrade it (in fact I've bought a second body, and will add more when used prices drop further). This is because I find that the quality it is capable of surpasses 35mm and to be blunt it is all too easy to enter technological overkill based on the numbers game. I have printed shots to 30in x 20in off the 1DS and 24/1.4 (a stunning lens with an extremely bright viewfinder image - another consideration?) and whilst I suppose that they could hold marginally more detail, you'd have to look pretty close. As has been commented here, the real problem with underwater wides is the dome optic. Currently, the bigger the dome the better the quality is a pretty reasonable statement. Perhaps manufacturers could look into seeing if a better alternative may be now built - I am trying to persuade a lens designer friend to look into the problem using lens design software, but this requires a lot of input which he can only do out of interest. One day.......!

[whitey]

Kasey. I think I see your point of confusion.

<{POST_SNAPBACK}>

I actually don't think Kasey is confused at all - I think he's got it exactly right.

Now I don't have the EF-S lens, but I understand that it produces '1:1' image size on the APS-C sensor (as compared with the size of the subject), correct?.

A lens that produces 1:1 (Canon 100mm, Tamron 90mm, not the Canon 50mm which is 1:2) on 35mm format will produce greater apparent magnification on a crop sensor camera. We all know of course that it's cropping rather than magnification that's happening here, the the fact remains that absolute size in mm of the image on the sensor will be different between the EF-S and the EF lens, ergo the degree of magnification is in fact different.