Jump to content

What is the Q's true resolution?


Recommended Posts

Advertisement (gone after registration)

Another newbie question. I realize that the Q's sensor is rated at 24MP resolution, but given the lens's rather large distortion, I am wondering what the final resolution is after correction. How many MP are lost, or did Leica build the sensor so that it has the full 24MP after correction?

 

Rob

Link to post
Share on other sites

Which means that after correcting for distortion and cropping, the image frame contains about 22MP.

 

Some camera RAW apps that see the file without the correction show the full size of the sensor to be slightly larger, as in 10 pixels or so each dimension - (I don't recall the exact pixel measurement), but it's not much. If you do a search, you might find the thread. ACR with the correction shows it to be 6000px x 4000px. In any event, it's 24MP.

Edited by beez
  • Like 1
Link to post
Share on other sites

Some camera RAW apps that see the file without the correction show the full size of the sensor to be slightly larger, as in 10 pixels or so each dimension - (I don't recall the exact pixel measurement), but it's not much. If you do a search, you might find the thread. ACR with the correction shows it to be 6000px x 4000px. In any event, it's 24MP.

 

Darktable/Digikam show 6016x4016 pixel. Anyway my impression is that when appplying the correction much more than 16 pixel are lost.

 

- TK

Link to post
Share on other sites

Advertisement (gone after registration)

... much more than 16 pixel are lost.

 

- TK

 

 

I believe the extra 16 are mostly for color and noise detection.  Digital color sensors have extra pixels at the borders so the de-mosaicing algorithms have the needed data to get the colors right.  They may also use some of the extra pixels for dark noise detection.

 

No other pixels are lost.   They may be adjusted.  But then every pixel is adjusted for color or you'd be looking at raw data.

Link to post
Share on other sites

Thanks very much for clearing this up. 

 

Rob

 

Hm, I am not really sure. I just see 6016*4016 = 24.16MP before Darktable/Digikam apply the lens correction. But honestly I am not an expert here...

 

Tnx

 

TK

 

Link to post
Share on other sites

I use Photo Ninja, and it shows the full dimensions as 6120px X 4016px. After distortion correction and conversion to tiff, the dimensions are 6016px X 4016px. So I think that the final image is, indeed, 24MP.

 

Rob

Link to post
Share on other sites

Distortion correction does NOT lose pixels, it just morphs them but remains the same image pixel count.

 

It's just stretching pixels out, so yes in theory you are losing pixels, but it's not costing you resolution based on pixel count.

Link to post
Share on other sites

Distortion correction does NOT lose pixels, it just morphs them but remains the same image pixel count.

 

It's just stretching pixels out, so yes in theory you are losing pixels, but it's not costing you resolution based on pixel count.

 

But after correcting for distortion, some cropping is required. That means losing pixels. Nevertheless, the final resolution is 24MP. By my calculations, the full sensor is about 24.6MP based on Photo Ninja's pixel count. It is possible that other programs that give different counts.

 

Rob

Link to post
Share on other sites

[...]the image size is 6016*4016, 24.16MP, so its not a 26MP sensor.

We should not confuse image size and sensor size. The image size stated as pixel count in the metadata and the photo cell count present on the physical sensor are different animals. The EXIF (metadata created by my Leica Q) states image width to be 6180 (not 6016) and image height 4016. Those (24.82 mega)image pixels may come from a sensor having a different cell count. We can't tell the number of photo cells on the sensor just by looking at image size values stated in metadata. The metadata (from my own Leica Q) also states active area 6000x4000. Sure, we can assume the Leica Q copies digitized values for every photo cell on the sensor to the dng-file and states that full count in the metadata imagewidth and imageheight, but I will not bet my life on that. I would not be surprised if there is clever math involved before each cell value land on the memory card as "raw" data. I am not at all sure raw is as raw these days as we tend to think.

 

Leica, the maker, could let us know the facts here, but until then I'll just second MickyW and go shooting.

Link to post
Share on other sites

RawDigger shows 6120x4016 pixels, including masked pixels, which are on the right side of the fame. These are covered and not hit by light, they are commonly used for black point and/or dark current detection. Without the masked area, the effective sensor size is 6016x4016. After the "baked in" (opcode) lens correction, 6000x4000 remains. What happens during correction? RawDigger shows the captured image without the correction, so it is easy to compare the final result with the "raw" capture. (see screenshots below)

 

First thing that is striking is the dark corners. The image circle of the lens clearly does not cover the sensor corners. There is also a lot of distortion. Because of this, indeed a lot of captured pixels, way more than 16, are lost during correction. This can be seen on the screenshots, notice the space between the left edge and the bus stop sign for example. The final output is only 16 pixels smaller than the captured image, so the captured data is stretched, or rather warped, progressively from the center outward to the edges. What happens is that the more you get towards the edges of the final picture, the more the pixels are extrapolated by "stretching". This can be clearly seen in scenes with high detail in the edge areas. Sharpness and small detail is simply not as good as in the center because of this.

 

It is hard to speak of "effective pixel count", but it is clear that the final image is based on far less captured pixels than the 6016x4016 sensor area.

 

Why is this? It is simply a trade-off. By using software correction, lens design can be smaller and simpler. For a camera like the Q, a lot can be said for that, size and weight matters. And the lens is still very good IMO. But this also means that the camera is less suitable for subject matter with relevant high edge detail, pixel peeping or printing large will certainly show the deterioration. If you care about that, use another camera and lens. But as I use the camera for street and reportage, I could not care less. I care for the small size and speedy operation.

 

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

  • Like 1
Link to post
Share on other sites

This can be clearly seen in scenes with high detail in the edge areas. Sharpness and small detail is simply not as good as in the center because of this. [...] this also means that the camera is less suitable for subject matter with relevant high-edge detail, pixel peeping or printing large will certainly show the deterioration.

You are exaggerating.

 

The software correction of chromatic aberration and geometric distortion indeed loses some, umm ... sharpness, even at the frame's center, and more so near the edges and corners. But in order to see the sharpness difference between the corrected and the uncorrected image, you'd have to look at higher than 100 % view. I'd say 200 % is minimum; 300 % or 400 % is better. That means the difference is minuscule. Insignificant for any practical intent or purpose ... such as large-size printing.

 

As a matter of fact, while the corrected image is a tad softer indeed, no actual detail gets lost. Every pixel-sized feature seen in the uncorrected image can also be found in the corrected image, only with a diminutively softer definition. The software correction acts like a very weak anti-aliasing filter (which in some cases actually makes the appearance more natural, less digital). When the corrected image gets sharpened by a higher amount than the uncorrected image then they'll look almost identical. For an image with good contrast and lots of detail, an input-sharpening amount of 10 in Camera Raw or Lightroom is about the maximum the uncorrected file can take without ending up oversharpened. Crank the amount up to 20 or 25 (ACR's/Lr's default value) in the corrected image, and you'll be hard-pressed to see any difference between the corrected and the uncorrected image with regard to perceived sharpness and detail rendition.

Link to post
Share on other sites

I actually subscribed to Sean Reid Reviews just to read his multiple reviews of the Leica Q. Sean specifically addresses the issue of software distortion correction (SDC), and he concludes that it does indeed result in some loss of resolution in the corners in comparison to uncorrected files. SDC involves two layers of interpolation: first in redrawing the distortion-corrected file and second in resizing the cropped image back up to 24MP. The loss of resolution is visible at 100% enlargement. But in his concluding paragraphs, Sean makes it clear that none of this has much impact on viewing images at normal sizes either on screen or in print. In his view, the Q produces outstanding, robust images, especially in the center. Overall, he rates it as one of the best fixed lens cameras that he has ever reviewed.

 

I don't perform fancy tests, but I have compared my new Q to my old RX1. There is no doubt in my mind that the Leica outperforms the Sony in almost every meaningful way, including IQ. I do not own an RX1RII, but aside from higher resolution, I wonder how images from the two cameras compare.

 

Rob

  • Like 1
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...