Quick 5DSR vs 5D4 image comparison

[John Sheehy]

well the detail in the 5DsR shots is real, nothing fake about it, the AA cancellation doesn't create detail...

The lack of sufficient low-pass filtering can create fake detail.

It puts real detail in the wrong place and orientation, it turns high frequency details into low-frequency ones, and is inconsistent, depending on luck of alignment.

 

[John Sheehy]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

One of the main points of an AA filter is that it cuts frequencies above the pixel frequency, which have the potential to mirror as low frequencies, much more than frequencies just below or at the pixel frequency. So, when you sharpen, the real frequencies get boosted into visibility while the mirrored ones are still strongly suppressed.

 

[Rick Knepper]

then I guess all the reviews are wrong regarding the sharpness and detail benefits of the 5DsR... see link and how quick the 5DIV falls out, but you want to tell me its IQ with the AA is a better 'representation' http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison

 

[marshwader]

then I guess all the reviews are wrong regarding the sharpness and detail benefits of the 5DsR... see link and how quick the 5DIV falls out, but you want to tell me its IQ with the AA is a better 'representation' http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison

 

[J A C S]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

 

[Jostian]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Also, the main thing is how poor the 5DIV is compared to the rest by virtue of it have an AA, in this context.

 

[J A C S]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Also, the main thing is how poor the 5DIV is compared to the rest by virtue of it have an AA, in this context.

Fewer mp (and upscaled), slightly backfocused, and no AA filter. If you want to isolate one factor, the AA filter only, compare the 5DS and the 5DSR in the studio scene.

 

[Victor Engel]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye. Also, the pattern in the eyes apparently is a pattern printed on the contact lenses. Were you referring to something else?

--
Victor Engel

 

[Rick Knepper]

then I guess all the reviews are wrong regarding the sharpness and detail benefits of the 5DsR... see link and how quick the 5DIV falls out, but you want to tell me its IQ with the AA is a better 'representation' http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison

 

[Victor Engel]

Have you (or anyone) taken DPR's Studio images of the 5DsR and 5Ds and compared a deconvolution sharpened 5Ds to an unsharpened 5DsR? a perfect way to compare AA vs -AA it would seem.

The review on this site does such a comparison, although I don't know what kind of sharpening was used.

 

[marshwader]

Traditionally, I've traded a small amount of false detail for a whole bunch of sharper detail starting with the D800E b/c I was 'promised' sharper images. No one really questioned that too much (except for moire) until the 5D4 made an appearance. Hmmm.

Have you (or anyone) taken DPR's Studio images of the 5DsR and 5Ds and compared a deconvolution sharpened 5Ds to an unsharpened 5DsR? a perfect way to compare AA vs -AA it would seem.

In the final analysis though, if I had bought a 5Ds instead, I would not feel that I had purchased an inferior device.

 

[J A C S]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

Were you referring to something else?

No, I am referring to the pattern itself. Unless it can move by itself, it is a false image on an actual pattern which looks like dots.

 

[John Sheehy]

The best cure for false detail is 100 MPs.

100MP FF is not a lot, especially if you want a reasonable result stripping out the red or blue RAW channel for a pure B&W image, or are shooting under a narrow-band LED that only registers in a useful manner under one filter color. We already have pixels much smaller than that, with the ability to adapt DSLR lenses, and they alias.

 

[Victor Engel]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

What animation? I didn't see anything animating.

Were you referring to something else?

No, I am referring to the pattern itself. Unless it can move by itself, it is a false image on an actual pattern which looks like dots.

 

[poppyjk]

I'm certainly not an expert on all this, but recently learned a little about 'mazing' as an artifact that results in a cross hatch pattern like the ones seen on the contact lenses in the D850 and A7RIII closeup eye photos. But maybe the cross hatch is actually on the contact lens -- not sure. Certainly less visible on the Fuji and Canon closeups.

Here is an excerpt from a short thread on 'mazing'. https://www.dpreview.com/forums/post/60333290

"This indeed is a demosiacing artifact called "mazing" and varies in pattern and nature somewhat depending on the demosiacing algorithm used. It usually is most apparent when there is high contrast very fine pattern, texture, or noise, which is close to the size of the individual cells of the Bayer matrix and arises due to the algorithm attempting to "connect the dots" between what it tries to see as strong edges when there is no real consistent pattern in the data.

You may want to ask someone like Iliah Borg (author of LibRaw, RawDigger, FPV) for some further explanations



View: original size

No EXIF data.

View: original size

I think this phenomenon might also be the reason for the infamous "worming" or the "painterly effect" (mostly manifesting itself in foliage and grass), and gets especially exacerbated if you use deconvolution sharpening together with non-optimized demosaicing. "

 

[J A C S]

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

What animation? I didn't see anything animating.

If there is no aliasing, a moving pattern is the only explanation.

 

[Victor Engel]

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

What animation? I didn't see anything animating.

If there is no aliasing, a moving pattern is the only explanation.

I understand that, but where is the moving pattern? I saw only stills.

 

[J A C S]

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

What animation? I didn't see anything animating.

If there is no aliasing, a moving pattern is the only explanation.

I understand that, but where is the moving pattern? I saw only stills.

The pattern has change orientation from shot to shot. Has it rotated?

 

[Victor Engel]

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye.

There are some spots on the contact lenses which allows you to see how much they are rotated (not much).

Also, the pattern in the eyes apparently is a pattern printed on the contact lenses.

An animated one? Apparently, it is aliasing.

What animation? I didn't see anything animating.

If there is no aliasing, a moving pattern is the only explanation.

I understand that, but where is the moving pattern? I saw only stills.

The pattern has change orientation from shot to shot. Has it rotated?

Obviously, I don't know where the images are. If you're referring to the various eye pictures taken with the different cameras, the slight change in angle of the grid pattern is easily explained by the natural rotation that occurs in human vision. I imagine contact lenses shift as well, but I have no data on that.

--
Victor Engel

 

[Victor Engel]

I think you should be careful in making those claims regarding the AA filter. The 5div has a lower sampling rate due to the lower number of MP. The AA filter, as I demonstrated in the links above attenuates contrast. So your MTF curve is attenuated at the higher wave numbers but contrast isn't cut to zero there, which would be the only way to destroy detail.

Also, the pattern the A7RIII displays there is wrong (as in fake detail).

how is it wrong? I had a look again http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiA.HTM#comparison and don't see anything wrong...

Look at the direction of the pattern in the D850 and the A7RIII shots - quite different. Also, you see the typical greek pattern associated with aliasing. Contact lenses of this type have small dots like pattern.

Compare the crops with the full frame. The images have obviously been rotated to present a normal looking eye. Also, the pattern in the eyes apparently is a pattern printed on the contact lenses. Were you referring to something else?