MTF Charts vs. Sharpness in the field

[neuroanatomist]

The insane thing about the Great Tit image is the appalling noise in it - and at the low iso. And it's not particularly sharp. If that's typical Sony, well forget it.

That was my reaction, as well. I saw the filenames with WhatsApp in them and was hoping that means that app handled the images poorly and something was lost in translation. If that’s native Sony output…ugly with a capital UG.

 

[Cryve]

The insane thing about the Great Tit image is the appalling noise in it - and at the low iso. And it's not particularly sharp. If that's typical Sony, well forget it. Here is a shot from an R5 at iso 5000 with the RF 100-500mm. And, one from yesterday afternoon which is a small crop from the RF 200-800mm at 800mm, which isn't the sharpest lens out there at there when at that focal length.


View attachment 214731
View attachment 214733

Great shots and great lighting. The r5 is an amazing camera.

The picture I uploaded are 200% and 300% crops.

Just wasnt happy with sony before and now i am. Im glad you are happy with canon.

 

[Cryve]

That was my reaction, as well. I saw the filenames with WhatsApp in them and was hoping that means that app handled the images poorly and something was lost in translation. If that’s native Sony output…ugly with a capital UG.

The whatsapp compression probably doesn't help but they are extremely heavy crops (200%/300%) with a 61mp sensor. Keep in mind that it means it's Zooming in way further into the picture than a 200% zoom in on a 45 mp sensor. It's just to show sharpness.

If I downscale to 45 mp or just view it not cropped the noise is as good as any other modern camera or at least comparable.

 

[Cryve]

I'm quite surprised that it's that sharp actually because It seems it's actually engineered for an even higher megapixel count.

I thought the megapixel race might be over, at least I don't need that many more personally.

Sony is in a tough spot here for the next model. If they increase the megapixel they will have to stick with 10 frames/second, and if they go with a fast readout sensor they probably have to go with same or lower megapixel for the next sensor iteration.

I feel canon is more balanced here with 20 feames/sec and 45 in the r5 and I don't even wanna know what the r5 ii will bring.

If sony goes for higher mp then they will be left in the dust regarding frames/sec by Canon and Nikon who do 20 frames/sec already. My a7r iv does some lousy 6 or so a second.

But if they go stacked or fast readout they can't stay true to the R series. R for resolution.

 

[EricN]

The Sony photos are noisier than I expected for 800 iso, but I think for some people the higher mp number is worth it as well as the global shutter/faster max shutter speed with the a9iii. Of course, we are too familiar with people wanting the third-party auto focus lenses, as well.
So, I can understand someone accepting more noise if any of those are priorities, and I'm glad @Cryve is happy now.

 

[AlanF]

The Sony photos are noisier than I expected for 800 iso, but I think for some people the higher mp number is worth it as well as the global shutter/faster max shutter speed with the a9iii. Of course, we are too familiar with people wanting the third-party auto focus lenses, as well.
So, I can understand someone accepting more noise if any of those are priorities, and I'm glad @Cryve is happy now.

The Great Tit photo was taken in APS-C mode on the Sony, which is 24 Mpx. The R7 has a 32 Mpx sensor, and doesn't have that noise at iso 800.

 

[EricN]

The Great Tit photo was taken in APS-C mode on the Sony, which is 24 Mpx. The R7 has a 32 Mpx sensor, and doesn't have that noise at iso 800.

That's a good point. I was wondering about it, but not having used an R7, had no idea.

 

[AlanF]

That's a good point. I was wondering about it, but not having used an R7, had no idea.

Here's a crop of a Nuthatch taken last week by wife with the R7 at iso 4000 (RF 100-400mm). It's straight out of DxO PL with. no added denoising by Topaz. I can just see the noise - it would be far more obvious using DPP4 etc (I recall you are interested in the denoising).

 

[EricN]

Here's a crop of a Nuthatch taken last week by wife with the R7 at iso 4000 (RF 100-400mm). It's straight out of DxO PL with. no added denoising by Topaz. I can just see the noise - it would be far more obvious using DPP4 etc (I recall you are interested in the denoising).

View attachment 214741

Wonderful photo! I'm quiet fond of images where we can see some change in depth of feild as with the tail end of the Nuthatch and rock in the corner.
It took me awhile to not feel like denoising looked too artificial.
I'm had mixed results with running my DxO PL images through Topaz. That is to say I thought some areas looked more natural and some looked worse.

After the other post, I was wondering about the Fuji xh2 sensor and the OM-1 sensor noise. I was looking at Claff's site and it'a interesting, but at least for me, some photos of charts would be better. I'm not sure if it's a group of people have already done this or if it's not reasonable to think about enough people agree about creating the same conditions for testing their cameras (same calibration and color charts, same lighting equipment and same room size and wall surface for keeping the ambient light the same plus I'm probably thinking of another variable).