Avoid 2x TC on A7Riv. Too much "empty" magnification!

[blue_skies]

I think that you sum it up well in your first post - using TCs degrades a lens quality, but unfamiliarity with the longer FL degrades an image faster.

Shooting 62Mp at 800mm handheld is also an art, and doing pixel peeping analysis may or may not be where it is at.

Adding a TC element to a lens never increases its optical quality. Sony must have known this by offering both the 1.4x and 2.0x strengths.

Just how does it compare with zooming in post? I'd get the impression that shooting the 1.4x TC in APS-C crop mode may yield better results.

I have used TCs in the past - they were fun, and compact, but the IQ never lived up to that of the original lens. From what I see, on the lower res (24Mp) this is a useful option when cropping is not always feasible. On the A7rIV this may be the other way around.

Suggestion, can you borrow a lens which reaches the FL without TC and compare the results with this lens? It may hint as to the testing method pitfalls. Try at least a (heavy) tripod when doing lens quality assessments.

 

[dcstep]

Be sure to look at 100%, full-screen.

FE 400mm f/2.8, plus FE 2.0x teleconverter:



FE 600mm f/4, plus FE 2.0x teleconverter:



For reference, here's the bare FE 600mm f/4 GM OSS:









I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

--
Dave

 

[bob13bob]

This is not good testing. If you truly want to test how good this combination works then you need to be a tripod, electronic shutter, wireless remote release after you are sure vibrations have died down completely. And shoot a planar target straight on, I can't even tell what is supposed to be in focus in your sample.

You don't need any of this. His testing should be how the lenses are actually used.

Northrup did tests on this. he found that TC degraded the image more than just cropping in pp. TC are another glass element, and they aren't going to make the image more clear..

for op's test, it would be nice to SEE a TC shot, then another shot cropped to same FL.

If he is going to blame physics or optics then he need to take his own technique out of the equation. To get the maximum out of the A7R IV with long lenses the technique must match the resolution.

And Tony Northrup tests a different lens. Not relevant in this case.

northrup tests variety of tcs on variety of bodies on a variety of lens. and shares some of his evidence

are you sure tihs isnt a case of i dont like the like the results,so i will nitpick the data. do you have any evidence to the contrary? it easy enough to prove us wrong, all you need is a tc tested on some lens.

these systrms are designed and engineered to be hand held. in testing, i e typically uses a high shutter speed and multiple shots to make sure thats eliminated.

Are you trying to generalize TN's results in that video to apply to all teleconverters and lenses or are you refering to something else?

There is a site which tests lenses and teleconverters with proper technique, you can check it out for yourself and decide if you find it worth it or not. Not all Sony lenses are tested there, not the 400GM for instance but here is an example with the 600GM:

https://www.the-digital-picture.com...eraComp=1175&SampleComp=0&FLIComp=1&APIComp=0

thank you for bringing some data to the table.

In that data, teh non TC picture is sharper than the TC. However, they seem to be moving the distance to the picture. If that is true, it's not a relevant test. We want to see a lense, eg 600 f4 cropped to 840, and then TCed to 840. way less relevant than the data I provided.

 

[bob13bob]

Be sure to look at 100%, full-screen.

FE 400mm f/2.8, plus FE 2.0x teleconverter:



FE 600mm f/4, plus FE 2.0x teleconverter:



For reference, here's the bare FE 600mm f/4 GM OSS:



I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

thanks for doing this.

i compared the two 600 shots, both 100 iso on a7r3. The 2x TC is much better than digitally cropping. see here for the comparison. I only could download the < 3mb jpeg files, but I'm still convinced by the comparison. It might be a little less contrast, but the imo the extra resolution makes up for it.

https://photos.app.goo.gl/M2th9y3DQZyNAg7x7

I stand corrected, for IQ, 2x TC is definitely better on some lens. From these results, I'm guessing that IQ will be more effected by the weakest link. for this prime on a7r3, it's the sensor resolution. for the sony 200-500, it would be the lens optics.

how do you find focus speed of a7r3 with TC usage vs non TC? Northrup states the AF on the 200-500 on the a7r3 and a73 takes quite a AF hit.

--
my equipment: a7iii. NATIVE: sony 50 1.8. samyang 35 2.8. ADAPTED: sigma mc11 adapter. canon 85mm 1.8. sigma (canon) 12-24 4-5.6. canon 70-300mm f/4-5.6 is ii.

 

[randomguy]

how do you find focus speed of a7r3 with TC usage vs non TC? Northrup states the AF on the 200-500 on the a7r3 and a73 takes quite a AF hit.

Assuming you mean the 200-600, it will take a massive hit on the A7R III because that camera does not support PDAF at smaller apertures than F8 so you get stuck with contrast detection. The A7III should however be able to handle it, you sure he said that one is also hit? The A9 however takes converters without a noticeable performance loss (except for the loss of light in low light of course), except maybe the 2x on the 200-600, haven't tested that for BIF.

--
Best Regards,
Morten Smedsrud
https://www.flickr.com/photos/154088944@N03/
https://500px.com/msmedsru

 

[bob13bob]

can you upload the full resolution files of dollar bills so i can eliminate the possibilty of difference being due to website compression.

 

[Rol Lei Nut]

I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

For those photographing dollar bills handheld with long teles + 2xTC, fine, I guess...

 

[Paul Barnard]

If he is going to blame physics or optics then he need to take his own technique out of the equation.

Perhaps you missed my earlier response where I said

"Maybe I could have got a slightly better handheld result than posted, but I can assure you that previous tests done on a tripod were identical. And all soft."

Yeah but just using a tripod is not enough, if you magnify the view while you have it on a tripod you will see how much and how long it shakes after you touch it. Even if there is a little wind it will shake badly on the tripod.

With the same lens, using the technique I described above and my A7RIII I get much better results than you posted even accounting for the resolution difference of the A7RIV and III. So I can't agree with your conclusion based on your sample.

 

[randomguy]

If he is going to blame physics or optics then he need to take his own technique out of the equation.

Perhaps you missed my earlier response where I said

"Maybe I could have got a slightly better handheld result than posted, but I can assure you that previous tests done on a tripod were identical. And all soft."

Yeah but just using a tripod is not enough, if you magnify the view while you have it on a tripod you will see how much and how long it shakes after you touch it. Even if there is a little wind it will shake badly on the tripod.

With the same lens, using the technique I described above and my A7RIII I get much better results than you posted even accounting for the resolution difference of the A7RIV and III. So I can't agree with your conclusion based on your sample.

your assumption is the lens degradation vs pixel pitch is linear. I would guess that isn’t the case and there is a distinct point at which the degradation becomes much worse quickly.

Comparisons against the A7Riii are meaningless. More relevant are comparisons with the APS-C sensors of similar pitch.

Lens resolution is independent of sensor resolution. Lenses do not suddenly become worse with higher resolution sensors.

 

[Beatsy]

OK, just to remove any doubt about my lens or TC not being good copies, I redid my previous tripod-based test. This time I used camera mounted flash (partially bounced, partially direct light - but identical for all images except power adjustments to match exposures). Flash range was around 1/16th to 1/8th power.

Distance to target was 12ft - a regularly-used range for small critters and birds (baited).

Well, I got a surprise - and a bit of egg on my face in the process. The 2x TC is capable of w-a-a-a-y sharper results than I thought, but still soft compared to no TC or the 1.4x (as we expect). It's actually quite good to confirm that with even *more* practice to improve my handheld technique I can squeeze even more performance out of this combo too. I love headroom :-D

Here's the test shots - insets are at 100% when images are viewed at "origanl size".


400GM




400GM+1.4xTC




400GM+2xTC

 

[Paul Barnard]

If he is going to blame physics or optics then he need to take his own technique out of the equation.

Perhaps you missed my earlier response where I said

"Maybe I could have got a slightly better handheld result than posted, but I can assure you that previous tests done on a tripod were identical. And all soft."

Yeah but just using a tripod is not enough, if you magnify the view while you have it on a tripod you will see how much and how long it shakes after you touch it. Even if there is a little wind it will shake badly on the tripod.

With the same lens, using the technique I described above and my A7RIII I get much better results than you posted even accounting for the resolution difference of the A7RIV and III. So I can't agree with your conclusion based on your sample.

your assumption is the lens degradation vs pixel pitch is linear. I would guess that isn’t the case and there is a distinct point at which the degradation becomes much worse quickly.

Comparisons against the A7Riii are meaningless. More relevant are comparisons with the APS-C sensors of similar pitch.

Lens resolution is independent of sensor resolution. Lenses do not suddenly become worse with higher resolution sensors.

 

[randomguy]

If he is going to blame physics or optics then he need to take his own technique out of the equation.

Perhaps you missed my earlier response where I said

"Maybe I could have got a slightly better handheld result than posted, but I can assure you that previous tests done on a tripod were identical. And all soft."

Yeah but just using a tripod is not enough, if you magnify the view while you have it on a tripod you will see how much and how long it shakes after you touch it. Even if there is a little wind it will shake badly on the tripod.

With the same lens, using the technique I described above and my A7RIII I get much better results than you posted even accounting for the resolution difference of the A7RIV and III. So I can't agree with your conclusion based on your sample.

your assumption is the lens degradation vs pixel pitch is linear. I would guess that isn’t the case and there is a distinct point at which the degradation becomes much worse quickly.

Comparisons against the A7Riii are meaningless. More relevant are comparisons with the APS-C sensors of similar pitch.

Lens resolution is independent of sensor resolution. Lenses do not suddenly become worse with higher resolution sensors.

The resolving power of the lens combined with the size of the sensor cells is what matters. So yes a sensor/lens combination can suddenly get worse. It’s nothing to do with resolution, though resolution is related to the things that do matter, it’s size/pitch of the sensor cells that matters. The sensor in the Riii has significantly bigger cells than the Riv or the latest ASP-C sensors. It’s all about the circle of confusion

Yes, if a lens has a 2pixel wide COC on a 24MP sensor then on a 96MP sensor same size it becomes 4 pixel wide. The CoC is due to the lens and does not change size regardless of sensor and thus can't suddenly grow faster than linear vs pixel pitch which you seem to claim in your previous post.

 

[dcstep]

thanks for doing this.

i compared the two 600 shots, both 100 iso on a7r3. The 2x TC is much better than digitally cropping. see here for the comparison. I only could download the < 3mb jpeg files, but I'm still convinced by the comparison. It might be a little less contrast, but the imo the extra resolution makes up for it.

https://photos.app.goo.gl/M2th9y3DQZyNAg7x7

I stand corrected, for IQ, 2x TC is definitely better on some lens. From these results, I'm guessing that IQ will be more effected by the weakest link. for this prime on a7r3, it's the sensor resolution. for the sony 200-500, it would be the lens optics.

how do you find focus speed of a7r3 with TC usage vs non TC? Northrup states the AF on the 200-500 on the a7r3 and a73 takes quite a AF hit.

Nice comparison.

Readers should understand that the 600mm f/4 is the primary tool of many bird-in-flight photographers. Even at 600mm, we're often cropping. A 50% to 80% crop is not unusual, so using a teleconverter to get more pixels on the subject is VERY useful. Taking a 1200mm shot and cropping it 10% (yes, you still might crop at that focal length) can be better than shooting at 600mm and cropping 90%, IF (that's a big "if") the teleconverter doesn't degrade the IQ too much.

My testing and Bob's further analysis shows that there is very little degradation with Sony's FE 2.0x teleconverter.

BTW, DPReview now has their set-up comparison shots up for the a7RIV. Look at those vs. the a9 and a7RIII. It's making me Jones of an a9II, with 30+mp, OR an a7x that'll match the a9's AF. The very best lens and teleconverter can't make up for OOF shots, with too low shutter speed.

 

[dcstep]

how do you find focus speed of a7r3 with TC usage vs non TC? Northrup states the AF on the 200-500 on the a7r3 and a73 takes quite a AF hit.

Assuming you mean the 200-600, it will take a massive hit on the A7R III because that camera does not support PDAF at smaller apertures than F8 so you get stuck with contrast detection. The A7III should however be able to handle it, you sure he said that one is also hit? The A9 however takes converters without a noticeable performance loss (except for the loss of light in low light of course), except maybe the 2x on the 200-600, haven't tested that for BIF.

I'd bet that the a9 works very well (I don't mean great) with the 200-600, plus 2.0x, BUT only in good light. It'll be okay for static shots in most light, but it'll get very sluggish and hunt too much in less than ideal light.

I shot a deer with my a9 and a bare 400/f2.8, AT ISO 64000. The focus was hunting like crazy, so I had to use the focus-lock buttons to hold focus while I took the shot. NO WAY I could have shot a BIF.







--
Dave

 

[dcstep]

can you upload the full resolution files of dollar bills so i can eliminate the possibilty of difference being due to website compression.

When you click on "Original Size", doesn't it show full rez?

Send me an email address and I'll send you the files via Dropbox

 

[dcstep]

I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

For those photographing dollar bills handheld with long teles + 2xTC, fine, I guess...

I used a tripod, in case I wasn't clear.

 

[dcstep]

OK, just to remove any doubt about my lens or TC not being good copies, I redid my previous tripod-based test. This time I used camera mounted flash (partially bounced, partially direct light - but identical for all images except power adjustments to match exposures). Flash range was around 1/16th to 1/8th power.

Distance to target was 12ft - a regularly-used range for small critters and birds (baited).

Well, I got a surprise - and a bit of egg on my face in the process. The 2x TC is capable of w-a-a-a-y sharper results than I thought, but still soft compared to no TC or the 1.4x (as we expect). It's actually quite good to confirm that with even *more* practice to improve my handheld technique I can squeeze even more performance out of this combo too. I love headroom :-D

Here's the test shots - insets are at 100% when images are viewed at "origanl size".

Thanks for going the extra mile and correcting your original conclusion.

It's too easy to bash the FE 2.0x teleconverter and too many people pile on and repeat the "factoid". YES, it does degrade the original, but is that enough for critical photographers to exclude it from usage for all circumstances? I think not, particularly for bird and bird-in-flight photographers.

 

[aSevenArr]

If you're planning to use (only) A7R4 with zooms or long primes on teleconverters, don't bother getting the 2x TC.

For pixel-level sharpness on A7Riv using a 2x TC (ignoring bayer matrix factors) a lens has to resolve the equivalent of 244 megapixels on FF! Sony claims "100-megapixel-ready" for GM lenses so clearly this is way out of spec and borne out by results. Here's a couple of 100% crops, one shot with A7Riv and a "naked" 400GM, the other with a 2x TC added. Each is the best result selected from several handheld shots. The 2x TC version isn't properly sharp anywhere - as expected from the numbers.


100% crop with 2x TC


100% crop, no TC

The 1.4x TC fares much better. When used on A7Riv it "only" requires the lens to resolve the equivalent of about 120 megapixels. This is still over the claimed capability, but not by much. Again, this is reflected in real-world results. Images taken with the 1.4x TC are acceptably sharp, though not *quite* as sharp as they are without the TC. Here's a snap taken this morning to prove the point - when viewed "original size" the inset is a 100% pixel-for-pixel crop.


1.4x TC (inset 100% when "viewed original size")

The above is all related to A7Riv only. The larger pixels of the A9 mean the issue is not so bad as the lens doesn't need to resolve anywhere near as much for pixel level sharpness on the 24MP sensor (96MP for 2x and 47MP for 1.4x). Hence why most TC users report good results on that camera.

A7Rii/iii fall somewhere between (lens needs to resolve 168MP and 82MP for 2x and 1.4x respectively). So the 2x doesn't look very good at all and the 1.4x is fine. Again - matched by real world results and reports.

Finally, a related observation. When I first used the 1.4x TC with my shiny new 400/2.8GM, I was disappointed with the results. Nothing was properly sharp. Turned out it wasn't the glass, just that I wasn't allowing for how much *more* steady my handheld technique needed to be at the longer FL. A little more concentration nailed sharp shots thereafter. I wonder if this could contribute to the apparent majority view that TCs perform poorly? If I hadn't figured out it was me at fault, I'd probably be another contributor to the "TCs are rubbish" meme now.

Thank you.... It’s an interesting write up but I knew (even from my 24MP A9 shots) that the 1.4X was just a teeny bit sharper than the 2X - you really don’t need a high megapixels A7rIV to draw that conclusion.

And yes camera shake is a problem for the sharpest results... I find IBIS and the built in lens stabilization is really very good but by no means perfect.

 

[Rol Lei Nut]

I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

For those photographing dollar bills handheld with long teles + 2xTC, fine, I guess...

I used a tripod, in case I wasn't clear.

I meant more that the results at short distances and longer ones are often different.

I long ago gave up testing longer lenses indoors, in order to get meaningful results (unless macro or close-up/portrait is the intended main use of said lenses).

 

[PWPhotography]

I hand hold almost all my field shots at these focal lengths, but typically at 1/2000-sec and higher. For testing and making declarative statements, I think that a tripod is required to get repeatable, meaningful results. The pixel difference may cause some deterioration, but I doubt that it's near so large with these lenses and teleconverters.

For those photographing dollar bills handheld with long teles + 2xTC, fine, I guess...

I used a tripod, in case I wasn't clear.

I meant more that the results at short distances and longer ones are often different.

I long ago gave up testing longer lenses indoors, in order to get meaningful results (unless macro or close-up/portrait is the intended main use of said lenses).

That is what I have said a few times but dcstep seems disagreed. For such long lenses, really the result is only meaningful if you shoot from distance, ideally at least 50x of FL. So for a 600mm FL lens, you need to shoot at least 30m away, basically reflect your usage in real world. Ideally setup a tripod outdoor to shoot on a big building/architecture from a distance.

I personally don't believe testing on a currency bill is that meaningful as they all appear sharp (because of aliasing) that also related to on how you align tripod properly.

For the same reason I don't take DPR or IR or TDP studio tests too seriously as they all taken from close distance. DPR real-world scene from the roof of the their Seattle HQ is much more meaningful and revealing in my opinion.

--
https://www.flickr.com/photos/55485085@N04/albums