[dtmateojr]

OK, please explain it to me like I'm a six year old, because I am, not for the first time, failing to understand.

I can shoot my 50mm lens on my APS-C body at f5.6 and obtain a specific image. I can also obtain, from the same shooting position, the same image (i.e. same exposure, FOV, and DOF) using a 75mm lens on my FF at f8. As it happens, my APS-C was set to ISO 100, which resulted in a shutter speed of 1/125th sec. My FF was also set to ISO 100, which resulted in a shutter speed of 1/60th sec. And that is that.

I now want to repeat the above, but the scene is darker, which results in my APS-C indicating a shutter speed of 1/30th, even after bumping ISO to1600. The the FF body now indicates a speed of 1/15th sec, also at ISO 1600. Under the circumstances, I am not sufficiently confident that this 1/15th SS is appropriate, and I wish for a minimum of 1/30th. I therefore bump my FF ISO to 3200, which magically results in the FF meter indicating a shutter speed of 1/30th, and I am where I want to be.

Now, if I were to view both of the above (second shoot) images at the same size, I understand DXO to be telling me that since the sensor tech level of my APS-C and FF is the same, I should pretty well end up with the same noise level. I believe there have been numerous posts demonstrating the lower noise level of FF using DXO methodology. But I understand you to be saying that such is not the case.

So where am I going wrong?

Your mistake is that you are equating image size (number of pixels) to REAL SNR (light gathering).

Downsampling an image increases SNR. The larger the image (not necessarily sensor) the better its downsampling advantage. This is why if you compare the noise performance of the D800 (36Mp) and D4 (16Mp) at a print size of 8Mp (8x10 DXO standard) they will come out equal. HOWEVER, as soon as you increase the print size to 16Mp, the D4 will come out superior to the D800. In FACT the D4 will come out superior to the D800 even at 36Mp (enlarged) print sizes.

Note that the D800 and D4 are both FF so total light is the same but we notice significant variations in SNR at different print sizes. The D4 beats the D800 at its own game (36Mp enlargements). The only time the D800 matches the D4 in performance is at smaller prints. Total light and image size do not correlate. They have absolutely nothing to do with each other. Do not use image size as basis for REAL SNR measurements. The FF comes out "better" than the APS-C not because it has gathered more light but simply because it has more pixels. Now manufacturers know this so they cram gazillions of pixels even on P&S cameras because nobody prints larger than 4x6, if the stupid consumers ever print at all. Yeah I'm exaggerating a bit but when was the last time you printed your photos?

My point is that, SNR comparisons between APS-C and FF by resampling and equating the result to light gathering is COMPLETELY WRONG. The method and conclusions are wrong even when comparing two FF (as per D800 vs D4), it's doubly wrong when comparing different formats.[COLOR="Silver"]

---

Last edited by Parallax; 07-23-2015 at 12:43 PM.

[clackers]

I was only talking about an increase in light gathering in terms of opening up the aperture more (or using a faster lens).

You did not.

I quoted you accurately.

You said amount of light and aperture as if they were the same thing - dare we say, "Equivalent?" And that's what people would have found you saying if they browsed this topic.

You then compounded your mistake by asking "so you are challenging the idea that additional light gathered will reduce DOF?"

I'm afraid slowing shutter speed will have no effect.

So if what you're saying now is a retraction/clarification, however you want to pitch it, fine.

May I also remind you that you did not point any errors in falconeye's equivalence formulas. It is understandable, because there are no errors

What?

It's a pointless polemic, what is anyone supposed to say?

It first of all trots out standard formulas to hook you in, then does a non sequitir with its conclusion that "It is a myth that APS-C cameras "crop the sweet spot" of the image field of lenses which cover a larger image circle or are made for a full registration distance."

You're easily impressed.

I'll forgive him his claim (made twice) that his procedure was empirical, because English is not his first language. It was a desk exercise at best, cherry picking a website. He fails to control variables (and admits it) in this convoluted thought experiment. Two cameras, two lenses, and somebody else's data, manipulated.

I suggest that you can do by yourself a much better experiment, valid because it actually demonstrates what you're trying to prove, and is empirical instead of just saying that it is.

Get a full frame camera, put a lens on it, and take a tripod stabilised brick wall shot with and without cropped mode. The wider the aperture and focal length, the clearer the effect will be.

That's it.

Develop the two RAW images yourself, because you don't want in-camera or computer software to apply any JPEG fixes (this may be harder in some recent cameras that do distortion and vignetting fixes to 'RAW' files anyway - IIRC, new m43 cameras are an example, looking up tables in the lenses themselves).

Check the corners for vignetting and loss of detail.

Or don't shoot at all, because we know in advance the result.

Instead, do what anyone can do from their desk and check the MTF of any lens ever made.

[Fogel70]

That's not what the original statement was about. It was about what I said in my previous post. That's why I said that perhaps you misunderstood dtmateojr when he first used the term, "projected image size." I'm guessing that Fogel70 misunderstood as well. Incidentally, just in case there is any question, it doesn't matter what system a lens was made for, at the same focal length, the projected image size will be the same (at least as long as it's mounted at the correct registration distance for the lens of course).

To call this "projected image size" is incorrect IMO, as that is what is used FI for a projector lens when changing distance or focal length the projected image changes (the size of the projected image change). What happens in a camera is rather that "projected image magnification" changes with focal length and the projected image size is constant depending on that sensor size the lens was designed for.

Edit:
After all, with different focal length used a different FOV is captured by the lens, and this is projected on the sensor. So it's the light entering/passing through the lens that varies with focal length, not the size of the image projected by the lens. As longer focal length captures narrower FOV, it will capture less of the available light, which explains why a lens with longer focal length projects a image with less intensity of light if using the same size on aperture as on a lens with shorter focal length, that capture more of the available light (wider FOV).
And this is why f-stop is used for exposure, as this will lead to constant illumination independent of focal length, as the size of the aperture increase with focal length to compensate for the lower amount of light captured by longer focal length.

[clackers]

not if changes the perspective...

Who knows what subcult of Equivalence you belong to, Osv.

If you agree that moving a FF camera closer to the subject in these number contests is a conman's trick, good on you.

But you are the guy who once claimed there were six Sigma lenses available for Sony APS-C, because you cleverly counted the black and silver versions of the same three models.

And that URL you included?

Joseph James is no scientist. I've read plenty of genuine optics papers as a former Materials Engineering student and Physics teacher.

AFAIK, he's a fashion photographer chucked out of the DPR forum in the past who did a diatribic blogpost some, like you, have bookmarked as a kind of holy relic.

The time I once spent reading it is time I won't get back, unfortunately.

It riled your friend Dtmateojr alright:


https://dtmateojr.wordpress.com/2014/09/28/debunking-equivalence/

---

Last edited by clackers; 07-16-2015 at 10:34 PM.

[audiobomber]

Downsampling an image increases SNR. The larger the image (not necessarily sensor) the better its downsampling advantage. This is why if you compare the noise performance of the D800 (36Mp) and D4 (16Mp) at a print size of 8Mp (8x10 DXO standard) they will come out equal. HOWEVER, as soon as you increase the print size to 16Mp, the D4 will come out superior to the D800. In FACT the D4 will come out superior to the D800 even at 36Mp (enlarged) print sizes.

Note that the D800 and D4 are both FF so total light is the same but we notice significant variations in SNR at different print sizes. The D4 beats the D800 at its own game (36Mp enlargements). The only time the D800 matches the D4 in performance is at smaller prints. ]

Got any proof of these FACTS?

[BrianR]

Oh, come on, Brian!

This is another example of confusing correlation with causation.

Opening up the aperture let in more light, but opening up the aperture also made the cone of light (however dark or bright) falling on the sensor broader.

It was the aperture that was the key.

It wasn't just any 'additional light' he was talking about, it was exactly the additional light from opening the aperture, which is exactly the additional light that widens the blur discs. It's a domino in the causation chain, wider aperture -> light on a new path -> larger blur discs -> less DoF. If you're not willing to say this additional light caused the wider DoF then we're now in a strange semantics argument that I'm just not getting.

You're taking his quote out of context and trying to imply he's saying any source of additional light would change the DoF. He wasn't and if it wasn't clear to you from his post it should be by now.

[clackers]

. If you're not willing to say this additional light caused the wider DoF

No, Brian, the causation is the aperture widening and the *effects* are additional light and increased DoF.

We will just have to agree to disagree if you think otherwise.

[CFWhitman]

To call this "projected image size" is incorrect IMO, as that is what is used FI for a projector lens when changing distance or focal length the projected image changes (the size of the projected image change). What happens in a camera is rather that "projected image magnification" changes with focal length and the projected image size is constant depending on that sensor size the lens was designed for.

Edit:
After all, with different focal length used a different FOV is captured by the lens, and this is projected on the sensor. So it's the light entering/passing through the lens that varies with focal length, not the size of the image projected by the lens. As longer focal length captures narrower FOV, it will capture less of the available light, which explains why a lens with longer focal length projects a image with less intensity of light if using the same size on aperture as on a lens with shorter focal length, that capture more of the available light (wider FOV).
And this is why f-stop is used for exposure, as this will lead to constant illumination independent of focal length, as the size of the aperture increase with focal length to compensate for the lower amount of light captured by longer focal length.

Yes, I understood your thinking, which is why I was explaining that this was a miscommunication/misunderstanding. I was just stating what was meant by dtmateojr in his original post. There was no point in arguing back and forth about the physics when the actual issue was just one of semantics.

[dtmateojr]

Got any proof of these FACTS?

Obviously you did not read the link I provided. Scared of simple math and data?

[CFWhitman]

It wasn't just any 'additional light' he was talking about, it was exactly the additional light from opening the aperture, which is exactly the additional light that widens the blur discs. It's a domino in the causation chain, wider aperture -> light on a new path -> larger blur discs -> less DoF. If you're not willing to say this additional light caused the wider DoF then we're now in a strange semantics argument that I'm just not getting.

You're taking his quote out of context and trying to imply he's saying any source of additional light would change the DoF. He wasn't and if it wasn't clear to you from his post it should be by now.

You're right that this is basically an argument over semantics. However, the problem with calling it "additional light" is that it's not the amount of light that makes the difference; it's where the light travels. You can have less light because of environmental changes in a picture at f/1.4 and ISO 400 than you do at f/2 and ISO 200, and the picture with less light will still have a shallower depth of field because the light that you do have is distributed a different way.

[normhead]

The pitfall of calling this additional light is clear. If you have two equal light sources, turn off one and then open you're aperture you have the same amount of light, but decreased DoF. Just do what everyone else does and say a wider aperture creates less DoF, and there are no exceptions.

SO why not do that? or we could discuss it for another couple of pages.

[audiobomber]

Obviously you did not read the link I provided. Scared of simple math and data?

The link is not evidentiary, as it is just a series of opinions posted by Some Guy On The Internet. I reject outright his claim that DXO Screen measurements are useful for any practical purpose. Images are printed and viewed at specific sizes. Comparing a 16mp camera at 16mp vs. a 36mp camera at 36mp is misguided and unfair.

I'm looking for your proof that a D4 image blown up to 36mp will look better than a D800 image. Resolution at 16mp on FF is severely limited vs. 36mp, your D4 image will be mushy in comparison to a D800 image. Have you never looked up lens comparisons on DXO?

• D4 with Nikon AF-S Nikkor 58mm f/1.4G, sharpness = 11mp
• D800 with Nikon AF-S Nikkor 58mm f/1.4G, sharpness = 18mp

Tests and reviews for the lens Nikon AF-S Nikkor 58mm f/1.4G mounted on Nikon D4 - DxOMark
Tests and reviews for the lens Nikon AF-S Nikkor 58mm f/1.4G mounted on Nikon D800 - DxOMark

Of course if you really want to show off the performance of higher megapixels, you leave off the blur filter, because finer resolution means less moiré.

• D810 with Nikon AF-S Nikkor 58mm f/1.4G, sharpness = 25mp

[Fogel70]

I reject outright his claim that DXO Screen measurements are useful for any practical purpose.

It can be useful as it's basically show pixel noise, and if comparing cameras with same pixel size it can be useful.

Fi like here where dtmateojr try to prove that Nikon D800 and D7000 have same low light performance as they have same per pixel SNR.
https://dtmateojr.wordpress.com/2014/05/19/megapixel-hallucinations/

[audiobomber]

It can be useful as it's basically show pixel noise, and if comparing cameras with same pixel size it can be useful.

Only if the sensors are the same size too. Even then, it won't tell you anything different than the normalized graphs.

[osv]

Who knows what subcult of Equivalence you belong to, Osv.

If you agree that moving a FF camera closer to the subject in these number contests is a conman's trick, good on you.

it's a real simple concept, pretty much everyone else gets it :-)

"DEFINITION OF EQUIVALENCE
Equivalent photos are photos of a given scene that have the:

Same Perspective

Same Framing

Same DOF, Diffraction, Total Amount of Light on the Sensor

Same Shutter Speed

Same Brightness

Same Display Dimensions"

Joseph James is no scientist. I've read plenty of genuine optics papers as a former Materials Engineering student and Physics teacher.

funny how both you and dematojr claim to have these advanced credentials?? but the first thing any schoolboy learns is how to reference credible sources, which you both failed at big time

---------- Post added 07-17-15 at 10:07 AM ----------

If you want the math and REAL data see here:

https://dtmateojr.wordpress.com/2014/05/19/megapixel-hallucinations/

linking to sources that are not credible doesn't prove anything.

"The problem really is that he(dtmateojr) will ignore absolutely anything that doesn't fit his pet theory. That includes just about all the practical evidence and correct application of the physics. There is no reason in any discussion with him, you just play it for the laughs." -bobn2
You do not understand f-stop part II: Photographic Science and Technology Forum: Digital Photography Review

bobn2 is the guy behind sensorgen.info:
Sensorgen Expired?: Photographic Science and Technology Forum: Digital Photography Review