Quantifying the Effects of Camera Motion on Image Quality

[Doppler9000]

Inspired by recent discussion about the 100RF, I wondered about a thought experiment, comparing:

A. 24x36mm, 60 MP camera with 5 stops of stabilization, 35mm lens

B. 33x44mm, 100 MP camera with no stabilization, 35mm equivalent lens

The cameras are identical, optically, etc.

Is there a way to think about adding motion to the cameras, incrementally, until the image quality (MTF?) of the two systems is equal, such that we could conclude something like “If there is more motion than X, camera A will produce a higher quality image?

 

[Crispy_Bee]

The difference between 100MP and 60MP isn't as big as it sounds:

11648×8736

vs

9504 x 6336

so that's 0,23 x 0,38 "pixels" per pixel you could "mess up" to make up the difference.

So even a slight softening (due to camera motion, vibration,...) will reduce the 100MP to well below 60MP.

Also: using a higher ISO to get a faster shutter speed will have a similar result when it comes to details vs noise.

There's a reason why IBIS is kind of a "standard requirement" for even lower res but especially high-res sensors as it would otherwise lead to constant frustrations and a lot of trashed images. Just try shooting without IBIS and without OS (this is still a thing) for a few days and you'll see why it's so valuable.

Honestly the GFX100RF really needs IBIS or at least(!) OS, as it's really made/designed to be cropped and what good is a high-res sensor you could crop if the images have camera shake? I guess they wanted to remedy that with the leaf shutter (and also reduce the overall size even further), but it's simply not enough to eradicate camera movement.

 

[Ap0ll0n]

Yes but. A shaken image is still better than no image, if the FF or crop camera is too big to carry. Plus, ibis is not always needed or doing much. So a similar question to ask is what % of the time ibis is important. For some it may very well be under 20%, or, it doesn't have to be 99% for this camera to make (a lot of) sense.

Great question posed though, perhaps someone can measure.

 

[rbf]

Yes but. A shaken image is still better than no image, if the FF or crop camera is too big to carry. Plus, ibis is not always needed or doing much. So a similar question to ask is what % of the time ibis is important. For some it may very well be under 20%, or, it doesn't have to be 99% for this camera to make (a lot of) sense.

Great question posed though, perhaps someone can measure.

Good question. I remember Jim mentioning that IBIS at wider focal lengths can't keep both the center and the corners sharp, where wider means focal lengths that are less than about the diagonal of the sensor(55mm for GFX), if I remember correctly. I've also run into this wider lenses on my X series where I left IBIS on for a shot that probably didn't need and got all kinds of wonky stuff going on in the corners and outer areas of the image. I am new to IBIS and its quirks. But if this is the case and the GFX100RF has a 35mm lens well below the threshold for IBIS to keep the whole image sharp then how valuable is it really? You can also raise the shutter speed to get it sharp as well(1/30, 1/60... etc).

 

[Nitnaros]

My take on measurement:

- use each camera handheld
- take a large # of images, 100+
- determine % of keepers / sharp

if you define according your liking a keeper % threshold, then you can compare the minimum shutter speed for each camera platforms make that threshold

….only for your technique, of course

I am wondering how much benefit the leaf shutter will generate compared to the (large) focal plane shutter of the 100S

 

[Crispy_Bee]

Yes but. A shaken image is still better than no image, if the FF or crop camera is too big to carry. Plus, ibis is not always needed or doing much. So a similar question to ask is what % of the time ibis is important. For some it may very well be under 20%, or, it doesn't have to be 99% for this camera to make (a lot of) sense.

Great question posed though, perhaps someone can measure.

Good question. I remember Jim mentioning that IBIS at wider focal lengths can't keep both the center and the corners sharp, where wider means focal lengths that are less than about the diagonal of the sensor(55mm for GFX), if I remember correctly. I've also run into this wider lenses on my X series where I left IBIS on for a shot that probably didn't need and got all kinds of wonky stuff going on in the corners and outer areas of the image. I am new to IBIS and its quirks. But if this is the case and the GFX100RF has a 35mm lens well below the threshold for IBIS to keep the whole image sharp then how valuable is it really? You can also raise the shutter speed to get it sharp as well(1/30, 1/60... etc).

Unless I'm mistaken that's for ultra wide angles, so below 20mm fullframe - I could be wrong but I think that's where it starts to get problematic. So for 35 on GFX it shouldn't be an issue and - most important of all - I see now reason why it can't be adjusted for the lens to "only" provide a certain degree of stabilisation - which would be better than nothing at all.

Besides, there's also OS / OSS / IS which would also be helpful. But having none at all and the only solution being "oh well, crank up the ISO and then crop the noisy images to your delight" is not really an ideal scenario for a 5.500 € camera.

 

[rbf]

Yes but. A shaken image is still better than no image, if the FF or crop camera is too big to carry. Plus, ibis is not always needed or doing much. So a similar question to ask is what % of the time ibis is important. For some it may very well be under 20%, or, it doesn't have to be 99% for this camera to make (a lot of) sense.

Great question posed though, perhaps someone can measure.

Good question. I remember Jim mentioning that IBIS at wider focal lengths can't keep both the center and the corners sharp, where wider means focal lengths that are less than about the diagonal of the sensor(55mm for GFX), if I remember correctly. I've also run into this wider lenses on my X series where I left IBIS on for a shot that probably didn't need and got all kinds of wonky stuff going on in the corners and outer areas of the image. I am new to IBIS and its quirks. But if this is the case and the GFX100RF has a 35mm lens well below the threshold for IBIS to keep the whole image sharp then how valuable is it really? You can also raise the shutter speed to get it sharp as well(1/30, 1/60... etc).

Unless I'm mistaken that's for ultra wide angles, so below 20mm fullframe - I could be wrong but I think that's where it starts to get problematic. So for 35 on GFX it shouldn't be an issue and - most important of all - I see now reason why it can't be adjusted for the lens to "only" provide a certain degree of stabilisation - which would be better than nothing at all.

Besides, there's also OS / OSS / IS which would also be helpful. But having none at all and the only solution being "oh well, crank up the ISO and then crop the noisy images to your delight" is not really an ideal scenario for a 5.500 € camera.

I remember asking Jim where the cutoff for that was and he said about the diagonal of the sensor, which is 55mm for the GFX. I've noticed the effect at 18mm and below on the X system. With the GFX I mostly shoot with it off and on a tripod, particularly with wider focal lengths.

 

[JimKasson]

Inspired by recent discussion about the 100RF, I wondered about a thought experiment, comparing:

A. 24x36mm, 60 MP camera with 5 stops of stabilization, 35mm lens

B. 33x44mm, 100 MP camera with no stabilization, 35mm equivalent lens

The cameras are identical, optically, etc.

Is there a way to think about adding motion to the cameras, incrementally, until the image quality (MTF?) of the two systems is equal, such that we could conclude something like “If there is more motion than X, camera A will produce a higher quality image?

Here is a methodology that could be extended to address your question:

Simulating motion blur: MTF50 and pictures - the last word

Yesterday, I reported on the results on simulations of a Zeiss Otus 55/1.4 at various f-stops, and showed how MTF-50 values translated into picture sharpness with simulated photographs. Today I'll do something similar, but varying motion blur instead of f-stop. I set up the sim with the pixel...

blog.kasson.com

 

[Rashkh]

Inspired by recent discussion about the 100RF, I wondered about a thought experiment, comparing:

A. 24x36mm, 60 MP camera with 5 stops of stabilization, 35mm lens

B. 33x44mm, 100 MP camera with no stabilization, 35mm equivalent lens

The cameras are identical, optically, etc.

Is there a way to think about adding motion to the cameras, incrementally, until the image quality (MTF?) of the two systems is equal, such that we could conclude something like “If there is more motion than X, camera A will produce a higher quality image?

Cognisys' rotary tables can resolve to .01° of movement and can be programmed. I think that would be precise enough to create repeatable tests.

 

[The Hat]

Honestly the GFX100RF really needs IBIS or at least(!) OS, as it's really made/designed to be cropped and what good is a high-res sensor you could crop if the images have camera shake? I guess they wanted to remedy that with the leaf shutter (and also reduce the overall size even further), but it's simply not enough to eradicate camera movement.

i do agree

even Fujifilm when introduced the IBIS for GFX series (https://www.fujifilm-x.com/en-gb/learning-centre/the-benefits-of-gfx-ibis/)

said "Stability becomes even more important with high-resolution sensors, like those in GFX cameras. As resolution rises, micro blur becomes evident."

"More useful is that IBIS improves resolution and sharpness at speeds like 1/50 sec, 1/80 sec or 1/100 sec and above, especially when handling is difficult. Fine detail is retained more readily, so images look like they were captured at much higher speeds"

 

[Crispy_Bee]

Honestly the GFX100RF really needs IBIS or at least(!) OS, as it's really made/designed to be cropped and what good is a high-res sensor you could crop if the images have camera shake? I guess they wanted to remedy that with the leaf shutter (and also reduce the overall size even further), but it's simply not enough to eradicate camera movement.

i do agree

even Fujifilm when introduced the IBIS for GFX series (https://www.fujifilm-x.com/en-gb/learning-centre/the-benefits-of-gfx-ibis/)

said "Stability becomes even more important with high-resolution sensors, like those in GFX cameras. As resolution rises, micro blur becomes evident."

"More useful is that IBIS improves resolution and sharpness at speeds like 1/50 sec, 1/80 sec or 1/100 sec and above, especially when handling is difficult. Fine detail is retained more readily, so images look like they were captured at much higher speeds"

The lack of IBIS also has another drawback:

No pixel-shift.. on a 5.500€ camera with what seems to be a really good lens that could certainly provide the image quality for a really nice moire-free 400MP image. Pity.

Imagine someone buys this camera thinking of taking architecture or interior shots on a tripod (which is what they'd have to) and then they don't even get 4x pixel shift... oof.

 

[Rashkh]

Honestly the GFX100RF really needs IBIS or at least(!) OS, as it's really made/designed to be cropped and what good is a high-res sensor you could crop if the images have camera shake? I guess they wanted to remedy that with the leaf shutter (and also reduce the overall size even further), but it's simply not enough to eradicate camera movement.

i do agree

even Fujifilm when introduced the IBIS for GFX series (https://www.fujifilm-x.com/en-gb/learning-centre/the-benefits-of-gfx-ibis/)

said "Stability becomes even more important with high-resolution sensors, like those in GFX cameras. As resolution rises, micro blur becomes evident."

"More useful is that IBIS improves resolution and sharpness at speeds like 1/50 sec, 1/80 sec or 1/100 sec and above, especially when handling is difficult. Fine detail is retained more readily, so images look like they were captured at much higher speeds"

The lack of IBIS also has another drawback:

No pixel-shift.. on a 5.500€ camera with what seems to be a really good lens that could certainly provide the image quality for a really nice moire-free 400MP image. Pity.

Imagine someone buys this camera thinking of taking architecture or interior shots on a tripod (which is what they'd have to) and then they don't even get 4x pixel shift... oof.

The pixel shift on my GFX100S is as close to useless as you can get. I was getting errors combining shots taken on a tripod in my living room often enough to completely give up on it.

 

[JimKasson]

The pixel shift on my GFX100S is as close to useless as you can get. I was getting errors combining shots taken on a tripod in my living room often enough to completely give up on it.

I've not had any issues with GFX 100x pixel shift indoors, and long as the tripod or camera stand is solid and I'm on slab on grade. With high res pixel shift that I've used, ground and building conducted vibration can be an issue.

 

[Rashkh]

The pixel shift on my GFX100S is as close to useless as you can get. I was getting errors combining shots taken on a tripod in my living room often enough to completely give up on it.

I've not had any issues with GFX 100x pixel shift indoors, and long as the tripod or camera stand is solid and I'm on slab on grade. With high res pixel shift that I've used, ground and building conducted vibration can be an issue.

I'm on the 15th floor so that's probably it. I suppose it's an unfair comparison but some of the mft systems can pixel shift while being handheld so I'd expect something less prone to error when mounted on a tripod.