Do teleconverters even make sense with high density sensors?

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution. If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two. And if we take into account that teleconverters aren't ideal, then cropping can produce even better results. Anybody wants to experiment?

forpetessake said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution. If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two. And if we take into account that teleconverters aren't ideal, then cropping can produce even better results. Anybody wants to experiment?

Click to expand...

IMO yes a 1.4x TC is better. I can do a test tomorrow for you with an 800mm lens with and without a 1.4x TC and from my experience, the 1.4x TC shots will be sharper than cropping without one on. I could be wrong, but don't think so Pete.

Experianced guess from thousands of shots tells me that even with a 1.4x TC, focal length beats that crop.

It has to be a bird shot though because I only go by feather details, on the same perch, same distance, etc and both setups wide open.

Would that work for you mate ??

I already know that the 800mm + 1.4x TC is one very sharp combination.......

http://www.dpreview.com/forums/post/51411080

Could cropping be just as good without the 1.4c TC, I think it will be sharper with the 1.4x TC. I've been proved worng many times over though before.

All the best Pete and if that works for you, I can certainly try it tomorrow mate.

Danny.

--
http://www.birdsinaction.com

My guess is that for a lot of teleconverters designed for use on film, your hunch might be correct - you are just magnifying an image that is not fully resolving the sensor anyway. Good quality TCs might be a different story.

Also bear in mind that at the pixel crop level, especially at faster speeds, the image can be noisy, limiting the effective resolution of the sensor. Downsampling can be a good way of reducing the visiblity of noise. On the other hand, your TC will lose you 1 or 2 stops of speed, increasing the noise, so the calculus gets complicated again.

Using the whole sensor as opposed to a crop of the sensor always yields better results. Teleconverters are still relevant.

You're assuming that those numbers are meaningful.


The mpix rating for the sony 500/4 is something like 13 or 16 megapixels. I can tell you that when I put my ancient minolta 600/4 on my a850 I get pixel level sharpness across the entire frame, which common sense tells me is 24mp of resolution.

When I put it on my nex 7, I get pixel level sharpness across its frame, from only the center of the lens' output. And when I put the cheap kenko 1.4x TC on the combo, I get pixel level sharpness across ITS frame, from only the center of the center of the frame, through even more glass to boot.

When I put the cheap Kenko 2x on that, then I start getting a little bit of fuzziness, but I'm still eeking out more info than enlarging/cropping afterwards.

Those numbers are meaningless masturbatory bullsh!t.

forpetessake said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution.

Click to expand...

I have to admit that I have absolutely no idea what you are saying here, and it makes no sense. Where do you get that using a 1.4x TC will cut your lens' resolution in half? That's not been even close to my experience. My 300mm F4 APO is razor sharp, and is exactly as razor sharp when shooting with a matched 1.4x TC - one woud have to inspect pixels almost individually at massive magnification to find a difference, and if they could, it certainly wouldn't be a loss of 50%.

forpetessake said:

If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two.

Click to expand...

I've been there and done that, and at the 1.4x TC level, the TC was significantly better than cropping and upsampling, each and every time, at every aperture setting. The 1.4X TC was always the better choice...not only offering sharper and better results compared to the upsampled version without the TC, but of course still being shot using the full sensor resolution and being resizeable, croppable, and higher overall resolution for printing or any other use. With 2x TCs, they can lose a noticeable amount of sharpness and then you might have to individually weigh whether you are getting enough advantage over cropping, but with a 1.4x TC, it's easily noticeable how much better the TC is and in no way does a good TC have anywhere near the kinds of losses you imply.

forpetessake said:

And if we take into account that teleconverters aren't ideal,

Click to expand...

I don't take that into account - my experience is quite the opposite - teleconverters are indeed ideal. If well made, matched to the lens they're paired with, and kept at 1.4X or less, the teleconverter loses virtually no measureable loss in sharpness, retains full sensor resolution output, and delivers more focal reach in a small overall package compared to the equivalent focal lens...I personally consider that quite ideal.

--

Justin
galleries: www.pbase.com/zackiedawg

I own a lot of teleconverters -- most came along with various old lens lost, but I've deliberately bought a few too. In the 1970s, I bought a 2X Vivitar to extend my 135mm... and found it quite disappointing. Via eBay, I bought a 6-element macro focus 2X (with a focus helical) and a Tamron 2X matched to my Tamron SP 90mm f/2.5 macro. Both of these are supposedly outstanding, but I wasn't all that impressed, and I have yet to meet a teleconverted lens that was good wide open.

As for the resolution argument against teleconverters, well, sampling an image at higher than Nyquist is not without benefit. So, even though the benefit is greater getting up to Nyquist, sampling a "soft" image with more pixels does improve the quality of the result somewhat. Further, the relationship between resolution of the base lens and the teleconverted pairing is not as straightforward as it at first seems. Still, resolution isn't the primary issue.

Teleconverters naturally magnify most image quality flaws and reduce contrast, but some lens + teleconverter pairings are pretty good stopped down 1-2 stops. My macro focus 2X and Tamron 2X are both in this category. Of course, that means a 2X isn't just cutting light by 2 stops, but requiring exposures with a 3-4 stops darker image, which easily can make camera shake an issue. Lesser magnifications tend to do better in this respect, but low magnifications (e.g., 1.4X teleconverters or "glass adapters" like the 1.2X used to put Canon FD lenses on Sony A-mount bodies) also have relatively flat elements that are prone to reflection problems.

The most positive thing I can say about teleconverters is that many pairings do really nice things to lens bokeh, essentially darkening the edges of the PSF. A fast 50 that has good bokeh can become a really great bokeh 100mm lens when paired with a decent 2X converter. The softness introduced by the teleconverter can actually help if you're trying to use the 50mm + 2X as a 100mm portrait lens, and I think that's why a lot of people bought them years ago....

In sum, I don't think teleconverters earn their keep as focal multipliers to replace a standard telephoto lens, although they can serve special purposes well. Ironically, I'm very impressed with the Speed Booster and Lens Turbo focal reducers. So, it's not really about adding glass degrading optical performance -- it really depends what that glass is doing.

DxOMark's MP resolution rating has really shown how stupid this obsession with MP is.

I'd much rather have an 8MP APS-C sensor with "huge", sensitive, color rendering pixels that give me speed and full detail than a smeary + slower 24MP one. Who is making 24MP prints on a regular basis?

forpetessake said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution. If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two. And if we take into account that teleconverters aren't ideal, then cropping can produce even better results. Anybody wants to experiment?

Click to expand...

The replies tell me that I need to clarify my original post.

First of all, my post is purely theoretical, the only teleconverter I have is very old and pretty bad, I already know the images taken with it are poor. That of itself neither proves nor disproves the supposition.

Secondly, I don't really know how DXO came with the numbers and whether those are applicable in this situation. Though my hunch based on experience tells me they are correct. Also, it's worth remembering that many good lenses have extinction resolution higher than the Nyquist, but it isn't particularly interesting, the MFT50 is traditionally been used as a measure of perceivable sharpness.

Thirdly, the result of adding teleconverter is increasing the image area proportionally to the conversion factor, thus reducing resolution and total light as a square of conversion factor. It's basically the opposite of what the focal reducers do. Cropping of an ideal sensor also reduces the resolution and the total light as a square of the crop factor. So if one takes an ideal teleconverter with zero aberrations and ideal sensor with infinite density the effects of the teleconverter and cropping with the same ratio should be exactly the same.

Fourthly, though it's true that higher pixel count always results in more details even when an image has much lower special frequencies, the practical results from 24MP and 12MP sensors sampling an equivalent 7MP image should be very close when printed 8x10 (DXO) or viewed on HDTV or monitor (most common case). Try resizing DPR test images from different sensors to HDTV resolution and you will not likely be able to distinguish 12MP from 24MP image. If that's already true about high resolution images, it's even more true when the original image contains only 7MP worth of information.

Now, neither teleconverters nor the sensors are ideal, so an experiment is in order.

forpetessake said:

forpetessake wrote:

forpetessake said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution. If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two. And if we take into account that teleconverters aren't ideal, then cropping can produce even better results. Anybody wants to experiment?

Click to expand...

The replies tell me that I need to clarify my original post.

First of all, my post is purely theoretical, the only teleconverter I have is very old and pretty bad, I already know the images taken with it are poor. That of itself neither proves nor disproves the supposition.

Secondly, I don't really know how DXO came with the numbers and whether those are applicable in this situation. Though my hunch based on experience tells me they are correct. Also, it's worth remembering that many good lenses have extinction resolution higher than the Nyquist, but it isn't particularly interesting, the MFT50 is traditionally been used as a measure of perceivable sharpness.

Thirdly, the result of adding teleconverter is increasing the image area proportionally to the conversion factor, thus reducing resolution and total light as a square of conversion factor. It's basically the opposite of what the focal reducers do. Cropping of an ideal sensor also reduces the resolution and the total light as a square of the crop factor. So if one takes an ideal teleconverter with zero aberrations and ideal sensor with infinite density the effects of the teleconverter and cropping with the same ratio should be exactly the same.

Fourthly, though it's true that higher pixel count always results in more details even when an image has much lower special frequencies, the practical results from 24MP and 12MP sensors sampling an equivalent 7MP image should be very close when printed 8x10 (DXO) or viewed on HDTV or monitor (most common case). Try resizing DPR test images from different sensors to HDTV resolution and you will not likely be able to distinguish 12MP from 24MP image. If that's already true about high resolution images, it's even more true when the original image contains only 7MP worth of information.

Now, neither teleconverters nor the sensors are ideal, so an experiment is in order.

Click to expand...

Even if all this is true, you can still benefit from a teleconverter by decreasing sensor noise.

Suppose for a lens that perfectly images 24MP of data onto a 24MP sensor. If you just want the central 6MP and crop, each pixel in the resulting 6MP image has 1 sensor pixel's worth of noise.

For a perfect teleconverter that enlarges the 6MP region over the entire sensor area, the resulting 24MP image has 4 sensor pixels for each image pixel. If you resize this to a 6MP image, any half-decent algorithm will average out the noise for the 4 sensor pixels, and you'll get a better image.

Obviously teleconverters aren't perfect, which is why you won't necessarily get a better image than with cropping, but I think your past experience with an admittedly-bad teleconverter might be rather rare for the good teleconverters you can get today.

First off, even back in film days, its always the case that the media out resolve the lens. There is always a question oof the lens not good enough but seriously lens can be made to deliver that much more, its just a case of diminishing return with significantly increased cost ( and also bulk and weight ).

And then there's the point of whether you need the reach and whether you got the shoot. The data as such the Dxo site put up certainly are helpful but its only telling that part of the story. A good fix focal together with a proper designed tele-converter can easily still image onto these high density sensor a perfectly fine image. The resolution part is only 1 area of imaging performance, there are others.

In the end also. A tele converter setup is perhaps still better than shoot and crop or not getting the shoot at all.

But then, there's another good reason why we ask for those long tele lens ( and those other focal length too ) , right ?

forpetessake said:

forpetessake wrote:

Now, neither teleconverters nor the sensors are ideal, so an experiment is in order.

Click to expand...

The experiment I ran as a DPReview challenge was:

http://www.dpreview.com/challenges/Challenge.aspx?ID=4717

A teleconverter shot won, but 7 of the top 10 were cropped. Not very conclusive, but cropping did have the edge.

The advantages of cropping include that it is easier, more flexible, and allows faster shutter speeds in the same lighting. Decades ago, film resolution wasn't very good. Thus, lenses often out-resolved the film and the drop in optical resolution using a teleconverter did not reduce captured resolution correspondingly -- thus, a teleconverter would record more detail in the selected area. Very few lenses out-resolve a 24MP NEX-7, so teleconverters have lost their primary benefi, while the benefits of cropping remain intact....

Incidentally, my challenge series is called "Technology Enabling Art," and here's my statement of its purpose:

Digital cameras and digital image processing constitute a new medium for art. This challenge series centers on making artistic images in ways that take advantage of the unique properties and capabilities of the medium. The goal is to help people discover that they can do things with cameras and image processing that they never realized were possible. If there is a little advanced technology introduced to push the medium even further, such as in-camera scripting using CHDK or new image processing algorithms on your PC, that's even better.

Franka T.L. said:

Franka T.L. wrote:

First off, even back in film days, its always the case that the media out resolve the lens.

Click to expand...

Very much disagree. Commercially-processed 135 film in the 1970s was only giving the equivalent of between 1.5MP and 3MP, mostly thanks to much higher noise levels (but also due to less-than-ideal processing). Even using high-resolution films like Panatomic-X, processed using Rodinal, controlling temp to 1/2 degree during processing, etc., I could barely match a 6MP digital image.

My collection of over 125 film-era lenses is filled with examples that handily out-resolve 10MP sensors at their optimal apertures. I don't have a single lens that clearly out-resolves my 24MP NEX-7 from corner to corner at anything approaching MTF50....

Hye said:

Hye wrote:

Hye said:

forpetessake wrote:

Hye said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution. If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two. And if we take into account that teleconverters aren't ideal, then cropping can produce even better results. Anybody wants to experiment?

Click to expand...

The replies tell me that I need to clarify my original post.

First of all, my post is purely theoretical, the only teleconverter I have is very old and pretty bad, I already know the images taken with it are poor. That of itself neither proves nor disproves the supposition.

Secondly, I don't really know how DXO came with the numbers and whether those are applicable in this situation. Though my hunch based on experience tells me they are correct. Also, it's worth remembering that many good lenses have extinction resolution higher than the Nyquist, but it isn't particularly interesting, the MFT50 is traditionally been used as a measure of perceivable sharpness.

Thirdly, the result of adding teleconverter is increasing the image area proportionally to the conversion factor, thus reducing resolution and total light as a square of conversion factor. It's basically the opposite of what the focal reducers do. Cropping of an ideal sensor also reduces the resolution and the total light as a square of the crop factor. So if one takes an ideal teleconverter with zero aberrations and ideal sensor with infinite density the effects of the teleconverter and cropping with the same ratio should be exactly the same.

Fourthly, though it's true that higher pixel count always results in more details even when an image has much lower special frequencies, the practical results from 24MP and 12MP sensors sampling an equivalent 7MP image should be very close when printed 8x10 (DXO) or viewed on HDTV or monitor (most common case). Try resizing DPR test images from different sensors to HDTV resolution and you will not likely be able to distinguish 12MP from 24MP image. If that's already true about high resolution images, it's even more true when the original image contains only 7MP worth of information.

Now, neither teleconverters nor the sensors are ideal, so an experiment is in order.

Click to expand...

Even if all this is true, you can still benefit from a teleconverter by decreasing sensor noise.

Suppose for a lens that perfectly images 24MP of data onto a 24MP sensor. If you just want the central 6MP and crop, each pixel in the resulting 6MP image has 1 sensor pixel's worth of noise.

For a perfect teleconverter that enlarges the 6MP region over the entire sensor area, the resulting 24MP image has 4 sensor pixels for each image pixel. If you resize this to a 6MP image, any half-decent algorithm will average out the noise for the 4 sensor pixels, and you'll get a better image.

Click to expand...

I understand why you would think so, but that interesting part is that the photon noise is the same in both cases. Ideal sensors would produce the same amount of noise whether using cropping or using an ideal teleconverter. Think about this way, the lens collects all light within its angle of view proportional to it's effective aperture and then the sensor collects exactly the same amount of light either from a crop or from a full frame. Crop will be brighter but area will be smaller, with teleconverter the area is greater but the brightness is lower. No matter what you do, how large or small sensors you use, for a given angle of view the S/N ratio is bound by the lens effective aperture. In practice sensors introduce read noise as well, and teleconverters lose light in the glass elements, hard to say which side wins.

ProfHankD said:

ProfHankD wrote:

ProfHankD said:

forpetessake wrote:

Now, neither teleconverters nor the sensors are ideal, so an experiment is in order.

Click to expand...

The experiment I ran as a DPReview challenge was:

http://www.dpreview.com/challenges/Challenge.aspx?ID=4717

A teleconverter shot won, but 7 of the top 10 were cropped. Not very conclusive, but cropping did have the edge.

Click to expand...

Interesting. I think the first image won not for technical reasons, rather because the subject is so adorable

ProfHankD said:

The advantages of cropping include that it is easier, more flexible, and allows faster shutter speeds in the same lighting. Decades ago, film resolution wasn't very good. Thus, lenses often out-resolved the film and the drop in optical resolution using a teleconverter did not reduce captured resolution correspondingly -- thus, a teleconverter would record more detail in the selected area. Very few lenses out-resolve a 24MP NEX-7, so teleconverters have lost their primary benefi, while the benefits of cropping remain intact....

Click to expand...

Yes, even if I called the death of teleconverters prematurely, their fate seems to be sealed.

ProfHankD said:

Incidentally, my challenge series is called "Technology Enabling Art," and here's my statement of its purpose:

Digital cameras and digital image processing constitute a new medium for art. This challenge series centers on making artistic images in ways that take advantage of the unique properties and capabilities of the medium. The goal is to help people discover that they can do things with cameras and image processing that they never realized were possible. If there is a little advanced technology introduced to push the medium even further, such as in-camera scripting using CHDK or new image processing algorithms on your PC, that's even better.

Click to expand...

I tested the Canon FD 1.4x TC on the Canon 500 and 800mm today. I'll post up results when I get a chance to go through them.

In short, the TC wins............. but not by a lot. Less than I thought.

Very little in it, but and its a big but, cropping gives you a smaller image at the end. You don't need to crop as much with the 1.4x TC so you end up with a larger image.

So lets say for arguments sake that both are equal in image quality, then the TC is still on top when it comes to more detail for the same sized image.

Its a bit like saying, throw the 800mm away and use the 500mm and just crop from the 500mm to make an 800mm shot.

That is not going to happen ;-)

All the best and both the lenses are darn sharp to start with.

Danny.

nzmacro said:

nzmacro wrote:

I tested the Canon FD 1.4x TC on the Canon 500 and 800mm today. I'll post up results when I get a chance to go through them.

Click to expand...

Thanks for doing the experiment, there are only few folks here with really good equipment to test the theory. As I mentioned I assumed the lens' native resolution is 14MP, based on DXO numbers, I can imagine the high end lenses can resolve beyond that. As well as the 24MP sensor resolution is actually lower due to demosaicing.

nzmacro said:

In short, the TC wins............. but not by a lot. Less than I thought.

Very little in it, but and its a big but, cropping gives you a smaller image at the end. You don't need to crop as much with the 1.4x TC so you end up with a larger image.

So lets say for arguments sake that both are equal in image quality, then the TC is still on top when it comes to more detail for the same sized image.

Click to expand...

Yes, more pixels is better, even if the scene doesn't contain high special frequencies, but for practical reasons it might be irrelevant if you have enough pixels. If people resize their images for HDTV (2MP) the differences due to 12 or 24MP sampling might be smaller than the other effects. On the other hand, if there aren't enough pixels in the crop, then the analog blur from TC probably looks better than the digital blur from upsizing.

nzmacro said:

Its a bit like saying, throw the 800mm away and use the 500mm and just crop from the 500mm to make an 800mm shot.

Click to expand...

No, that's different. Lenses are designed to keep the image circle as optimal as possible thus keeping the resolution high. The TC is very suboptimal solution, it's just spreading the image over a larger area thus lowering the resolution, its advantage is based on the assumption that the lens has a lot higher resolution than the recording media.

nzmacro said:

That is not going to happen ;-)

All the best and both the lenses are darn sharp to start with.

Danny.

--
http://www.birdsinaction.com

Click to expand...

Okay first off I don't take resolution tests, I know nothing about graphs (and would like to keep it that way), not interested in lines per mm, what does interest me is birds I'm very basic and both feet are on the ground. What I do have a fair idea about is long fast lenses, technique and yep, birds. ;-)

This could well get very boring and at the end of the day, its all only in my opinion.

All images were treated exactly the same with LR, no deviation from one file to the next. "Previous" was used after the file before it. All were sharpened and adjusted, hey they are RAW after all, but all exactly the same.

All camera setting were wide open, the only change was the shutter speed to compensate for the 1.4x TC being mounted. So the benefit should go (key word "should") to the naked lens without the 1.4x TC which gives a faster shutter speed. My tripod does not move when locked, its heavy and solid. 2 second timer was used.

Here we go and totally using the magnified view for each shot to check for feather details before the click !!.


Wide out and not something anyone would normally take at that distance


So adding the 1.4x TC


So this is fairly easy to see in LR, details are lacking, but at that range I'm not surprised. Looking at the pine cones the focus is pretty much bang on, could be a little heat shimmer at that distance, no surprise


Okay, obviously with the 1.4x TC there is more details and you can zoom in more. Still not something I would accept though ;-)

Trust me, that focus is pretty bang on in both of those and without the TC as said, its a faster shutter speed. Be interesting to see what would happen if the lens was stopped down a couple of stops. That might all change.

If you cropped the first shot to the same as the second shot, you would never see it from that range with those subjects.

So lets close in to normal type shooting and where the NEX-7 shines IMO, cropping with birds and feathers


So without the TC at a normal shooting distance for me this one although I can get a lot closer even with the 500mm , but this was about the crop.


So a crop form the above shot with no TC. Here's where it changes. That crop is a lot smaller than a crop from the 1.4x TC added. That's pretty darn good for that crop without the TC .


At this range all of a sudden the 1.4x TC shows the difference in magnification and pulls the subjects in more.


Bang, that's why i wanted a NEX-7. So that crop, although the same ratio crop (within reason), is a larger image than a crop from without the TC

So rather than having an image cropped to the same size, it all comes down to how much detail is in the crop to start with.

No I don't normally crop to those extremes, but both ways are pretty sharp, with or without. The only difference is you don't need to crop the 1.4x TC as much as being without it. So it still pulls in the details.

There's not really a lot in it at the end of day. The thing is, this might well all change from the 800 to the 300 and 500. Optically they are all different. I also prefer to shoot without the 1.4x TC ;-)

Bird shooters need to crop and sometimes fairly heavy, so we get to see the results of pixel peeping and noise in every single shot. Noise does come into this as well. The more we crop, the more noise shows up and that's why I never shoot at over ISO 400 and ISO 800 tops !!

All the best and if you want the RAW's Pete. let me know. Not a problem. You might like to crop and re-size to suit what you want. I doubt I've done that properly. I'm just basic Pete.

Danny.

--
http://www.birdsinaction.com

Thanks Danny for yet more wonderful bird pictures, and for the comparisons. And, thanks to all contributors in this fascinating thread - it is bookmarked here.

zackiedawg said:

zackiedawg wrote:

zackiedawg said:

forpetessake wrote:

Looking at the DXO testing of APS-C lenses, one can notice that even the best primes can pull no more than 14MP of equivalent resolution. That means that a lens with an ideal 1.4x teleconverter won't produce more than 7MP of equivalent resolution.

Click to expand...

I have to admit that I have absolutely no idea what you are saying here, and it makes no sense. Where do you get that using a 1.4x TC will cut your lens' resolution in half? That's not been even close to my experience. My 300mm F4 APO is razor sharp, and is exactly as razor sharp when shooting with a matched 1.4x TC - one woud have to inspect pixels almost individually at massive magnification to find a difference, and if they could, it certainly wouldn't be a loss of 50%.

zackiedawg said:

If that lens is used with high density sensor like NEX-7, then we have a situation of sampling 7Mp with 24MP of the sensor. If we simply crop the original image to the same size we would get the same 7MP of the lens sampled with 12MP of the sensor. I doubt there will be any perceivable difference between the two.

Click to expand...

I've been there and done that, and at the 1.4x TC level, the TC was significantly better than cropping and upsampling, each and every time, at every aperture setting. The 1.4X TC was always the better choice...not only offering sharper and better results compared to the upsampled version without the TC, but of course still being shot using the full sensor resolution and being resizeable, croppable, and higher overall resolution for printing or any other use. With 2x TCs, they can lose a noticeable amount of sharpness and then you might have to individually weigh whether you are getting enough advantage over cropping, but with a 1.4x TC, it's easily noticeable how much better the TC is and in no way does a good TC have anywhere near the kinds of losses you imply.

zackiedawg said:

And if we take into account that teleconverters aren't ideal,

Click to expand...

I don't take that into account - my experience is quite the opposite - teleconverters are indeed ideal. If well made, matched to the lens they're paired with, and kept at 1.4X or less, the teleconverter loses virtually no measureable loss in sharpness, retains full sensor resolution output, and delivers more focal reach in a small overall package compared to the equivalent focal lens...I personally consider that quite ideal.

--

Justin
galleries: www.pbase.com/zackiedawg

Click to expand...

If I can just go back to an old post of mine and the comments from Justin at the time. I went straight back there because I remembered what he said at the time. It stuck in my mind for sure............

"and STILL have even more room to crop. I'd say if you can handle getting them in the frame with the TC and that focal length, then it's well worth using the TC more."

This is it. You have with the 1.4x TC, as long as its good quality, a larger image to start with compared to cropping it. If the TC is good quality that simply = more details.

Both images can still be cropped, but only one will give you a larger image to work with.

Cropping increases noise significantly, a 1.4x TC wins in that area. Sure you loose shutter speed, but noise is a bit harder to deal with. Bird and nature shooters need to crop, the extra noise we don't need.

Degradation does come into it, that's optical formulas for ya and can't be avoided, but IMO, its still better than cropping where you loose anyway as well.

All the best.

Danny.

--