What's going on here?

DXO uses their own lens corrections. The sensor in the camera is somewhat larger than what the mFTs cameras use. DXO uses those pixels beyond the normal mFTs image when their lens correction is greater than the manufacturers specifications. Thus some DXO images will have more pixels than the default mFTs image.

drj3 said:

DXO uses their own lens corrections. The sensor in the camera is somewhat larger than what the mFTs cameras use. DXO uses those pixels beyond the normal mFTs image when their lens correction is greater than the manufacturers specifications. Thus some DXO images will have more pixels than the default mFTs image.

Click to expand...

That would explain it. But the pixels that olympus does not use is more than just a few. I shrunk the SOOC image until the difference between the two (RHS) was nearly zero. The ruler at the sides shows about 0.3" per side smaller, which works out to be about 100 pixels on each side of the sensor not used in the in-camera developer.



Seems like a lot.

David

Significant benefit to using DxO, in my experience.

12mm of course gives a lot of barrel distortion in a zoom that's digitally corrected, and DxO does specific lens-camera modules for pretty much any combo.

Oly's 12mm corrections look like these:




corrected 12mm jpeg




Uncorrected ORF

DxO is reclaiming more of the uncorrected image file. The Photo Lab distortion menu allows switching to manual and picking from barrel, pincushion and fisheye tools.

Rick

Skeeterbytes said:

Significant benefit to using DxO, in my experience.

12mm of course gives a lot of barrel distortion in a zoom that's digitally corrected, and DxO does specific lens-camera modules for pretty much any combo.

Oly's 12mm corrections look like these:

DxO is reclaiming more of the uncorrected image file. The Photo Lab distortion menu allows switching to manual and picking from barrel, pincushion and fisheye tools.

Rick

Click to expand...

I also noticed that I can turn off cropping. Doing that, the long side of the resulting image goes from 5184 to 5327 pixels. Oddly, the short dimension doesn't change.

Three versions here - all at 25% size. DxO crop on -LHS, DxO crop off Center, SOOC -RHS



Is there some advantage to knowing the image size is constant? I'd prefer that I get the biggest image possible. I can do that by setting crop to OFF, but why isn't that the default?

David

DavidWright2010 said:

Is there some advantage to knowing the image size is constant? I'd prefer that I get the biggest image possible. I can do that by setting crop to OFF, but why isn't that the default?

Click to expand...

The image dimensions and crop of straight-out-of-camera (and out-of-Workspace) images are as specified and as intended by the manufacturer. When raw conversion software yields more pixels, your image may have a different crop (affecting composition) and the pixels outside the manufacturer's original frame may show higher levels of distortion and aberration.

DavidWright2010 said:

drj3 said:

DXO uses their own lens corrections. The sensor in the camera is somewhat larger than what the mFTs cameras use. DXO uses those pixels beyond the normal mFTs image when their lens correction is greater than the manufacturers specifications. Thus some DXO images will have more pixels than the default mFTs image.

Click to expand...

That would explain it. But the pixels that olympus does not use is more than just a few. I shrunk the SOOC image until the difference between the two (RHS) was nearly zero. The ruler at the sides shows about 0.3" per side smaller, which works out to be about 100 pixels on each side of the sensor not used in the in-camera developer.



Seems like a lot. ...

Click to expand...

If you have not already done, view the RAW image without any corrections.

Below are JPG/ RAW images from an Panasonic FZ1000 from FZ1000 Review:


Panasonic FZ1000 WA RAW image without any corrections; saved as JPG for viewing.


Panasonic FZ1000 SOOC JPG. You can see how much the in-camera JPG processing is cropping the image.


Simple graphic I did to show approx. crop the in-camera JPG processing is doing..

Below the RAW image PP with DxO PhotoLab Elite Suite and DxO ViewPoint for Perspective and Volume corrections,



Cheers,
Jon

The camera specifications states the number of effective pixels is 20 MP and the number of sensor photo detectors as 22 MP. Some software for development of raw files will develop the unused sensor photo detectors, as stated by others. I discovered this many years ago when comparing the straight processing of different pieces of software to that from Olympus Viewer (now Workspace). There are colour differences, sharpness differences, and differences in the number of pixels in the processed images along with differences in the correction of lens deficiencies. After that, I used Viewer (or later Workspace) to export tiffs for further processing in my processor of choice. Fewer pixels but I liked the colour performance better and the images were often sharper. Some commentors on the posted results of the time said the Olympus program had more default sharpening applied. I had the in camera sharpening turned off but I don't know if Viewer applied sharpening anyway.

I tried DXO for development but never liked the default colour results and gave up on it. I last tried it about one version ago but still did not like the default colour. I have not tried Adobe in years but last time I tried it the results were noticeably less sharp, straight up.

Sometimes the extra pixels are good though and I develop in software that will use them. Choices are good!

Andrew

Agree with what you posted,

But it's quite common with wide angle zooms (to keep size/ cost down) that the lens image circle does not cover the entire sensor surface as per images I posted in my previous post .

Hence the in-camera PP crops image to exclude the areas at corners/ sides where lens' image circle does not cover the sensor.

Jon_T said:

Agree with what you posted,

But it's quite common with wide angle zooms (to keep size/ cost down) that the lens image circle does not cover the entire sensor surface as per images I posted in my previous post .

Hence the in-camera PP crops image to exclude the areas at corners/ sides where lens' image circle does not cover the sensor.

Click to expand...

Probably specific to fixed-lens cameras with rear elements extending into the light box, such as the LXs. Whether the LX variable aspect ratio feature plays a part might be a parallel consideration.

I'd be surprised if an ILC WA lens image circle did not cover the format's design dimensions (18x13.5mm) and that's where the extra sensor dimensions might come into play.

A couple GH models have oversize sensors and variable aspect ratios, but those are the exception to the m4/3 rule.

Rick

Skeeterbytes said:

Jon_T said:

Agree with what you posted,

But it's quite common with wide angle zooms (to keep size/ cost down) that the lens image circle does not cover the entire sensor surface as per images I posted in my previous post .

Hence the in-camera PP crops image to exclude the areas at corners/ sides where lens' image circle does not cover the sensor.

Click to expand...

Probably specific to fixed-lens cameras with rear elements extending into the light box,...

Click to expand...

Nope.

Similar issue with Canon EF and EF-S lenses.

Some of the new RF lenses (e.g., RF 24-240mm) rely heavy on software corrections.

That's precisely why I asked the OP if he looked at an uncorrected RAW image to verify.

Awhile back I used the Olympus 4/3rds DSLRs, and participated in this forum.

Had the E-500, E-510, E-3, and the E-620.









;-)

Jon_T said:

Skeeterbytes said:

Jon_T said:

Agree with what you posted,

But it's quite common with wide angle zooms (to keep size/ cost down) that the lens image circle does not cover the entire sensor surface as per images I posted in my previous post .

Hence the in-camera PP crops image to exclude the areas at corners/ sides where lens' image circle does not cover the sensor.

Click to expand...

Probably specific to fixed-lens cameras with rear elements extending into the light box,...

Click to expand...

Nope.

Similar issue with Canon EF and EF-S lenses.

Some of the new RF lenses (e.g., RF 24-240mm) rely heavy on software corrections.

That's precisely why I asked the OP if he looked at an uncorrected RAW image to verify.

Awhile back I used the Olympus 4/3rds DSLRs, and participated in this forum.

Had the E-500, E-510, E-3, and the E-620.





;-)

Click to expand...

What am I looking for, here? Corners look intact FWIW.

Skeeterbytes said:

Jon_T said:

Skeeterbytes said:

Jon_T said:

Agree with what you posted,

But it's quite common with wide angle zooms (to keep size/ cost down) that the lens image circle does not cover the entire sensor surface as per images I posted in my previous post .

Hence the in-camera PP crops image to exclude the areas at corners/ sides where lens' image circle does not cover the sensor.

Click to expand...

Probably specific to fixed-lens cameras with rear elements extending into the light box,...

Click to expand...

Nope.

Similar issue with Canon EF and EF-S lenses.

Some of the new RF lenses (e.g., RF 24-240mm) rely heavy on software corrections.

That's precisely why I asked the OP if he looked at an uncorrected RAW image to verify.

Awhile back I used the Olympus 4/3rds DSLRs, and participated in this forum.

Had the E-500, E-510, E-3, and the E-620.





;-)

Click to expand...

What am I looking for, here?

Click to expand...

Just post some of my E-620 pics to go with my comment. ;-)

Skeeterbytes said:

...Corners look intact FWIW.

Click to expand...

Of course.

As I clearly explained/ shown in my previous post need to view "unedited" RAW images for what I was speaking of as a "possible" cause to the OP's first post question.

Obviously SOOC JPGS or PP RAW images 'should' not show the effects of an lens image circle not fully covering the sensor. :-|