leerob wrote:according to Ken Rockwell, unless you have very good lenses. Does thids mean that lenses development lags behind sensor development. I'm talking about D7100 here.It's not "too much" but for the average camera user that KR is talking to, they will not see a significant difference between a D7000 and a D7100.  There are many reasons to buy a D7100 over a D7000 but resolution alone is not a big one. To squeeze the most detail out of the D7100 you're going to need a good quality lens, and you're going to need to have the camera on a tripod and use mirror lock up... and in most cases, unless you're printing really large, the average user isn't going to see much difference.  I have them both, I shoot them both side by side, there isn't that much difference but if I'm only shooting one camera, it's the D7100.

J C Brown wrote:mosswings wrote:It's well known that it takes a bit more than 2.4 pixels to resolve one line pair in a Bayer digital sensor owing to the Bayer Color Filter Array...only a panchromatic sensor would have the potential of reaching 2000 LP/PH, or 4000 LPH. This is where 80% of that 1.5 factor comes from.As to 85lp/mm, it's interesting to note that the oft-criticized Nikon 18-200 yields 2600 LW/PH at 50mm, f5.6 when tested on a D7000, or 86 lp/mm, and that includes the resolution of the sensor! All this says is that lenses are still quite good compared to the resolving capability of the sensors applied to them, and what we are seeing in the complaints department is mostly small things. Not night and day.From my reading of the specification in the DPR review the D7000 has 3264 pixels in a sensor height of 15.6 mm. If my calculations are correct that means that a resolution of 2600 LW/PH corresponds to a line width of 3264/2600 = 1.26 pixels and a resolution of 1300/15.6 = 83.3 lp/mm.As a digital image consists entirely of one pixel high squares it is clearly impossible for it to accurately represent a group of lines or spaces each of which has a width of 1.26 pixels. Only lines and spaces which are integer multiples of a pixel height can be represented accurately.If for a monochrome image it is assumed that the lines are parallel to the axes of the sensor then depending on its position in relation to the edges of a row of pixels a single 1.26 pixel wide black line may be recorded as a 2 pixel wide line in which one row of pixels is black and the adjacent row is 26% grey, a 2 pixel wide line in which the adjacent rows are of two different shades of grey or of a 2 pixel wide line in which both rows are of the same 63% shade of grey. Similar reasoning may be applied to the resulting nominally white spaces between the lines.That in my opinion is the main cause for the significant differences between the resolution values assessed from images of tapered line test charts, as discussed inhttp://www.dpreview.com/forums/post/41438422.I agree fully with your views about the quality of lenses in relation to the resolution of digital sensors. With the exception of low quality lenses unless it is adversely affected by aberrations at large apertures, diffraction at small apertures, noise at high ISO and/or blurring due to camera shake the resolution of a digital camera is in my opinion dominated by the resolution of the sensor and the subsequent in camera processing rather than by the quality of the lens.JimmyJC, the numbers come from photozone.de, which uses Imatest, and I rounded off their results for discussion purposes.  They mention that their numbers should not be used with extreme precision, as sharpening algorithms and other rendering issues can affect the number.But we also have to remember that what is actually being measured is a contrast ratio (MTF 50, 30, etc. at a particular lighting condition), which is a continuous function.  There's a lot of interpolation going on in the measurement and rendering, so we can get a result that doesn't square with the pixelation of the sensor.The important thing as well to remember here is that it takes a line PAIR - a dark and a light one - to define an edge, and the 2.52 pixels that you get from this calculation are perfectly in line with the resolution reduction expected for Bayer sensing.  Even if 2.52 pixels doesn't make practical sense.

In any case I think that D7100 is more advanced than many cameras used by great master photographers of the past. Let concentrate on on this.

mosswings wrote:J C Brown wrote:mosswings wrote:It's well known that it takes a bit more than 2.4 pixels to resolve one line pair in a Bayer digital sensor owing to the Bayer Color Filter Array...only a panchromatic sensor would have the potential of reaching 2000 LP/PH, or 4000 LPH. This is where 80% of that 1.5 factor comes from.As to 85lp/mm, it's interesting to note that the oft-criticized Nikon 18-200 yields 2600 LW/PH at 50mm, f5.6 when tested on a D7000, or 86 lp/mm, and that includes the resolution of the sensor! All this says is that lenses are still quite good compared to the resolving capability of the sensors applied to them, and what we are seeing in the complaints department is mostly small things. Not night and day.From my reading of the specification in the DPR review the D7000 has 3264 pixels in a sensor height of 15.6 mm. If my calculations are correct that means that a resolution of 2600 LW/PH corresponds to a line width of 3264/2600 = 1.26 pixels and a resolution of 1300/15.6 = 83.3 lp/mm.As a digital image consists entirely of one pixel high squares it is clearly impossible for it to accurately represent a group of lines or spaces each of which has a width of 1.26 pixels. Only lines and spaces which are integer multiples of a pixel height can be represented accurately.If for a monochrome image it is assumed that the lines are parallel to the axes of the sensor then depending on its position in relation to the edges of a row of pixels a single 1.26 pixel wide black line may be recorded as a 2 pixel wide line in which one row of pixels is black and the adjacent row is 26% grey, a 2 pixel wide line in which the adjacent rows are of two different shades of grey or of a 2 pixel wide line in which both rows are of the same 63% shade of grey. Similar reasoning may be applied to the resulting nominally white spaces between the lines.That in my opinion is the main cause for the significant differences between the resolution values assessed from images of tapered line test charts, as discussed inhttp://www.dpreview.com/forums/post/41438422.I agree fully with your views about the quality of lenses in relation to the resolution of digital sensors. With the exception of low quality lenses unless it is adversely affected by aberrations at large apertures, diffraction at small apertures, noise at high ISO and/or blurring due to camera shake the resolution of a digital camera is in my opinion dominated by the resolution of the sensor and the subsequent in camera processing rather than by the quality of the lens.JimmyJC, the numbers come from photozone.de, which uses Imatest, and I rounded off their results for discussion purposes. They mention that their numbers should not be used with extreme precision, as sharpening algorithms and other rendering issues can affect the number.But we also have to remember that what is actually being measured is a contrast ratio (MTF 50, 30, etc. at a particular lighting condition), which is a continuous function. There's a lot of interpolation going on in the measurement and rendering, so we can get a result that doesn't square with the pixelation of the sensor.The important thing as well to remember here is that it takes a line PAIR - a dark and a light one - to define an edge, and the 2.52 pixels that you get from this calculation are perfectly in line with the resolution reduction expected for Bayer sensing. Even if 2.52 pixels doesn't make practical sense.mosswings,Thanks for your comments and for taking the time to provide the above explanation.You may be interested in having a close look at the following image of the test chart described in the link in my earlier post. It was taken with my Nikon D7100 and 16-85 zoom lens mounted on a tripod. As will be seen from the EXIF data it was taken at 1/25 sec. at  F/4 with a focal length of 16 mm. Apart from using PSE to remove a slight blue cast from a lossless JPEG crop saved using FSV it has not been altered.In the above image the Es in the 1.4 and 1.5 pixel rows correspond to resolutions of 2857 and 2667 LPH respectively.I hope that you will find it of some interest.Jimmy

leerob wrote:according to Ken Rockwell, unless you have very good lenses. Does thids mean that lenses development lags behind sensor development. I'm talking about D7100 here.Actually any lens will perform better on the D7100 than on a lower pixed camera like the D7000. That's the case also when comparing the D800 x D610. You can see that in the measurement figures for D7000 x D7100 with same lens:http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7000__680http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7100__865

J C Brown wrote:mosswings wrote:J C Brown wrote:mosswings wrote:It's well known that it takes a bit more than 2.4 pixels to resolve one line pair in a Bayer digital sensor owing to the Bayer Color Filter Array...only a panchromatic sensor would have the potential of reaching 2000 LP/PH, or 4000 LPH. This is where 80% of that 1.5 factor comes from.As to 85lp/mm, it's interesting to note that the oft-criticized Nikon 18-200 yields 2600 LW/PH at 50mm, f5.6 when tested on a D7000, or 86 lp/mm, and that includes the resolution of the sensor! All this says is that lenses are still quite good compared to the resolving capability of the sensors applied to them, and what we are seeing in the complaints department is mostly small things. Not night and day.From my reading of the specification in the DPR review the D7000 has 3264 pixels in a sensor height of 15.6 mm. If my calculations are correct that means that a resolution of 2600 LW/PH corresponds to a line width of 3264/2600 = 1.26 pixels and a resolution of 1300/15.6 = 83.3 lp/mm.As a digital image consists entirely of one pixel high squares it is clearly impossible for it to accurately represent a group of lines or spaces each of which has a width of 1.26 pixels. Only lines and spaces which are integer multiples of a pixel height can be represented accurately.If for a monochrome image it is assumed that the lines are parallel to the axes of the sensor then depending on its position in relation to the edges of a row of pixels a single 1.26 pixel wide black line may be recorded as a 2 pixel wide line in which one row of pixels is black and the adjacent row is 26% grey, a 2 pixel wide line in which the adjacent rows are of two different shades of grey or of a 2 pixel wide line in which both rows are of the same 63% shade of grey. Similar reasoning may be applied to the resulting nominally white spaces between the lines.That in my opinion is the main cause for the significant differences between the resolution values assessed from images of tapered line test charts, as discussed inhttp://www.dpreview.com/forums/post/41438422.I agree fully with your views about the quality of lenses in relation to the resolution of digital sensors. With the exception of low quality lenses unless it is adversely affected by aberrations at large apertures, diffraction at small apertures, noise at high ISO and/or blurring due to camera shake the resolution of a digital camera is in my opinion dominated by the resolution of the sensor and the subsequent in camera processing rather than by the quality of the lens.JimmyJC, the numbers come from photozone.de, which uses Imatest, and I rounded off their results for discussion purposes. They mention that their numbers should not be used with extreme precision, as sharpening algorithms and other rendering issues can affect the number.But we also have to remember that what is actually being measured is a contrast ratio (MTF 50, 30, etc. at a particular lighting condition), which is a continuous function. There's a lot of interpolation going on in the measurement and rendering, so we can get a result that doesn't square with the pixelation of the sensor.The important thing as well to remember here is that it takes a line PAIR - a dark and a light one - to define an edge, and the 2.52 pixels that you get from this calculation are perfectly in line with the resolution reduction expected for Bayer sensing. Even if 2.52 pixels doesn't make practical sense.mosswings,Thanks for your comments and for taking the time to provide the above explanation.You may be interested in having a close look at the following image of the test chart described in the link in my earlier post. It was taken with my Nikon D7100 and 16-85 zoom lens mounted on a tripod. As will be seen from the EXIF data it was taken at 1/25 sec. at F/4 with a focal length of 16 mm. Apart from using PSE to remove a slight blue cast from a lossless JPEG crop saved using FSV it has not been altered.In the above image the Es in the 1.4 and 1.5 pixel rows correspond to resolutions of 2857 and 2667 LPH respectively.I hope that you will find it of some interest.JimmyI'd say 1.4 is minimum acceptable visually, which, interestingly enough, is what photozone.de reports the D7000 LP/PH resolution for the 16-85 is: 2832. It would interesting to test what your lens does on a D7000. Lots of factors here: visually acceptable vs. objectively determined contrast ratio; creeping moire as we approach the aliasing limit; actual lens limitations - which may be finally showing up at the D7100 resolution level.

mosswings wrote:J C Brown wrote:mosswings wrote:J C Brown wrote:mosswings wrote:It's well known that it takes a bit more than 2.4 pixels to resolve one line pair in a Bayer digital sensor owing to the Bayer Color Filter Array...only a panchromatic sensor would have the potential of reaching 2000 LP/PH, or 4000 LPH. This is where 80% of that 1.5 factor comes from.As to 85lp/mm, it's interesting to note that the oft-criticized Nikon 18-200 yields 2600 LW/PH at 50mm, f5.6 when tested on a D7000, or 86 lp/mm, and that includes the resolution of the sensor! All this says is that lenses are still quite good compared to the resolving capability of the sensors applied to them, and what we are seeing in the complaints department is mostly small things. Not night and day.From my reading of the specification in the DPR review the D7000 has 3264 pixels in a sensor height of 15.6 mm. If my calculations are correct that means that a resolution of 2600 LW/PH corresponds to a line width of 3264/2600 = 1.26 pixels and a resolution of 1300/15.6 = 83.3 lp/mm.As a digital image consists entirely of one pixel high squares it is clearly impossible for it to accurately represent a group of lines or spaces each of which has a width of 1.26 pixels. Only lines and spaces which are integer multiples of a pixel height can be represented accurately.If for a monochrome image it is assumed that the lines are parallel to the axes of the sensor then depending on its position in relation to the edges of a row of pixels a single 1.26 pixel wide black line may be recorded as a 2 pixel wide line in which one row of pixels is black and the adjacent row is 26% grey, a 2 pixel wide line in which the adjacent rows are of two different shades of grey or of a 2 pixel wide line in which both rows are of the same 63% shade of grey. Similar reasoning may be applied to the resulting nominally white spaces between the lines.That in my opinion is the main cause for the significant differences between the resolution values assessed from images of tapered line test charts, as discussed inhttp://www.dpreview.com/forums/post/41438422.I agree fully with your views about the quality of lenses in relation to the resolution of digital sensors. With the exception of low quality lenses unless it is adversely affected by aberrations at large apertures, diffraction at small apertures, noise at high ISO and/or blurring due to camera shake the resolution of a digital camera is in my opinion dominated by the resolution of the sensor and the subsequent in camera processing rather than by the quality of the lens.JimmyJC, the numbers come from photozone.de, which uses Imatest, and I rounded off their results for discussion purposes. They mention that their numbers should not be used with extreme precision, as sharpening algorithms and other rendering issues can affect the number.But we also have to remember that what is actually being measured is a contrast ratio (MTF 50, 30, etc. at a particular lighting condition), which is a continuous function. There's a lot of interpolation going on in the measurement and rendering, so we can get a result that doesn't square with the pixelation of the sensor.The important thing as well to remember here is that it takes a line PAIR - a dark and a light one - to define an edge, and the 2.52 pixels that you get from this calculation are perfectly in line with the resolution reduction expected for Bayer sensing. Even if 2.52 pixels doesn't make practical sense.mosswings,Thanks for your comments and for taking the time to provide the above explanation.You may be interested in having a close look at the following image of the test chart described in the link in my earlier post. It was taken with my Nikon D7100 and 16-85 zoom lens mounted on a tripod. As will be seen from the EXIF data it was taken at 1/25 sec. at F/4 with a focal length of 16 mm. Apart from using PSE to remove a slight blue cast from a lossless JPEG crop saved using FSV it has not been altered.In the above image the Es in the 1.4 and 1.5 pixel rows correspond to resolutions of 2857 and 2667 LPH respectively.I hope that you will find it of some interest.JimmyI'd say 1.4 is minimum acceptable visually, which, interestingly enough, is what photozone.de reports the D7000 LP/PH resolution for the 16-85 is: 2832. It would interesting to test what your lens does on a D7000. Lots of factors here: visually acceptable vs. objectively determined contrast ratio; creeping moire as we approach the aliasing limit; actual lens limitations - which may be finally showing up at the D7100 resolution level.Thanks for your further interesting comments and for your continued interest. I agree with your assessment of 1.4 as being the minimum that is visually acceptable.As you will see from its file number that test image was one of first few that I recorded with my D7100. Having used Panasonic FZ20 and FZ50 bridge cameras since October 2004 the D7100, my first DSLR, was a completely new experience for me. Although I’ve picked up a lot of valuable information and useful tips from the helpful members of this forum I’m still fairly close to the bottom of a steep learning curve.I was pleased to read that you think it would be interesting to see what my lens would do on a D7000. For several years I’ve been trying with limited success to encourage members of the Panasonic Compact Camera Forum to use my charts to assess and compare the performance of their cameras and TCs. The small number who have used them say that they have found them relatively easy to use and easier to interpret than an image of the tapered lines of an ISO chart.As you will see from my responses to John Miles I’ve recently constructed an improved version of my test chart which has been developed to allow reasonably accurate assessment and comparison of cameras and lenses of all brands and include the influence of several of the factors you mention.See:http://www.dpreview.com/forums/post/52980923andhttp://www.dpreview.com/forums/post/52988984As you will see from the following PSE screen capture by using the Nikon 35 mm F/1.8 lens I bought in December with Live View and the D7100’s built-in flash I’ve been able to record an image of the latest version of my chart in which three single pixel lines separated by two single pixels spaces can be fairly easily recognised at several positions in the 1.0 pixel row.Crop from 100 ISO JPEG taken at F/5.6 and showing 1.0 to 1.4 pixels rows of black lines with 1.0 pixel lines and spaces indicated in red. I hope you will find the above image of some interest.Jimmy

J C Brown wrote:I was pleased to read that you think it would be interesting to see what my lens would do on a D7000. For several years I’ve been trying with limited success to encourage members of the Panasonic Compact Camera Forum to use my charts to assess and compare the performance of their cameras and TCs. The small number who have used them say that they have found them relatively easy to use and easier to interpret than an image of the tapered lines of an ISO chart.As you will see from my responses to John Miles I’ve recently constructed an improved version of my test chart which has been developed to allow reasonably accurate assessment and comparison of cameras and lenses of all brands and include the influence of several of the factors you mention.See:http://www.dpreview.com/forums/post/52980923andhttp://www.dpreview.com/forums/post/52988984As you will see from the following PSE screen capture by using the Nikon 35 mm F/1.8 lens I bought in December with Live View and the D7100’s built-in flash I’ve been able to record an image of the latest version of my chart in which three single pixel lines separated by two single pixels spaces can be fairly easily recognised at several positions in the 1.0 pixel row.Crop from 100 ISO JPEG taken at F/5.6 and showing 1.0 to 1.4 pixels rows of black lines with 1.0 pixel lines and spaces indicated in red.I hope you will find the above image of some interest.JimmyJC -I presume your improved version is in monochrome with RGB and CMYK inking as per your post in the referenced threads?You are testing with perhaps the sharpest tool in the DX lens shed at this time (the FX Nikkors are better), but clearly the contrast has turned pretty much to mush at the 1.0 pixel level and the spaces between bars of the E are fairly evenly split between dark, light, and mid-grey. Note that you still have to climb back to about 1.3 before the pattern favors a 3 dark / 2 light sequence, and it's back to 1.4 before I'd say it is clearly that. And again, referencing photozone.de's tests of this lens on the D7000, we get 2737 LW/PH, and 2800 at f4, the peak for this lens. Conclusion: the 35mm looks to be a touch sharper than the 16-85, but tests at the same conditions would have to be run to be definitive.I can understand why users of your E chart would find it more satisfying than the tilted-edge Imatest pattern; your chart is designed for human viewing.  Basically, what you've done is discretized the hyperbolic wedge test on the classic ISO 12233 test chart and turned the triple bars on the 1951 USAF High-Resolution test chart into a phalanx of shrinking letters.  I do like it for its eye-chart friendliness, but computers like the tilted-edge pattern better.

You can ignore KR, but you would be wrongApproximate Linear Relationships of "Theoretical Cameras"Approximate pixel density(in pixels per linear centimetre)Pixel density in pixels per linear centimetre = image width in pixelsdivided bywidth of sensor in centimetres24 mp APS-C =2564.1(6000 / 2.34)36 mp FF =2042.5(7353 / 3.60)Pixel Density Advantage:24 mp APS-C is approximately 25.5% greater than 36 mp FFMore densely packed sensor sites will show up more flaws in a lens, that is just a fact, and I can see that KR knows this.

Zenjitsuman wrote:You can ignore KR, but you would be wrongYou can ignore what I wrote but you would be wrong..Approximate Linear Relationships of "Theoretical Cameras"Approximate pixel density(in pixels per linear centimetre)Pixel density in pixels per linear centimetre = image width in pixelsdivided bywidth of sensor in centimetres24 mp APS-C =2564.1(6000 / 2.34)36 mp FF =2042.5(7353 / 3.60)Pixel Density Advantage:24 mp APS-C is approximately 25.5% greater than 36 mp FFMore densely packed sensor sites will show up more flaws in a lens, that is just a fact, and I can see that KR knows this.I agree and have written many times describing the higher pixel density of 24mp APS-C sensors. Ken knows this, you know this and I know this. What you apparently don't know is why KR thinks it's necessary to reel in gullible rubes by misleading them with provocative statements. But if you really do understand what KR is all about and still think that he is deserving, take the KR test and prove it by your actions, not by your words. His ever growing family will appreciate it.https://www.paypal.com/xclick/business=paypal@kenrockwell.com&item_name=KenRockwell.com&no_note=1&tax=0¤cy_code=USD

mosswings wrote:J C Brown wrote:I was pleased to read that you think it would be interesting to see what my lens would do on a D7000. For several years I’ve been trying with limited success to encourage members of the Panasonic Compact Camera Forum to use my charts to assess and compare the performance of their cameras and TCs. The small number who have used them say that they have found them relatively easy to use and easier to interpret than an image of the tapered lines of an ISO chart.As you will see from my responses to John Miles I’ve recently constructed an improved version of my test chart which has been developed to allow reasonably accurate assessment and comparison of cameras and lenses of all brands and include the influence of several of the factors you mention.See:http://www.dpreview.com/forums/post/52980923andhttp://www.dpreview.com/forums/post/52988984As you will see from the following PSE screen capture by using the Nikon 35 mm F/1.8 lens I bought in December with Live View and the D7100’s built-in flash I’ve been able to record an image of the latest version of my chart in which three single pixel lines separated by two single pixels spaces can be fairly easily recognised at several positions in the 1.0 pixel row.Crop from 100 ISO JPEG taken at F/5.6 and showing 1.0 to 1.4 pixels rows of black lines with 1.0 pixel lines and spaces indicated in red.I hope you will find the above image of some interest.JimmyJC -I presume your improved version is in monochrome with RGB and CMYK inking as per your post in the referenced threads?You are testing with perhaps the sharpest tool in the DX lens shed at this time (the FX Nikkors are better), but clearly the contrast has turned pretty much to mush at the 1.0 pixel level and the spaces between bars of the E are fairly evenly split between dark, light, and mid-grey. Note that you still have to climb back to about 1.3 before the pattern favors a 3 dark / 2 light sequence, and it's back to 1.4 before I'd say it is clearly that. And again, referencing photozone.de's tests of this lens on the D7000, we get 2737 LW/PH, and 2800 at f4, the peak for this lens. Conclusion: the 35mm looks to be a touch sharper than the 16-85, but tests at the same conditions would have to be run to be definitive.I can understand why users of your E chart would find it more satisfying than the tilted-edge Imatest pattern; your chart is designed for human viewing. Basically, what you've done is discretized the hyperbolic wedge test on the classic ISO 12233 test chart and turned the triple bars on the 1951 USAF High-Resolution test chart into a phalanx of shrinking letters. I do like it for its eye-chart friendliness, but computers like the tilted-edge pattern better.Good to know all these.Thank you!

Just pointing out that KR isn't automatically wrong.   KR is fairly up on Nikon gear past and presenthey most people on the web are selling something and I let everyone make their spiel doesn't mean I buy it.

mosswings wrote:J C Brown wrote:I was pleased to read that you think it would be interesting to see what my lens would do on a D7000. For several years I’ve been trying with limited success to encourage members of the Panasonic Compact Camera Forum to use my charts to assess and compare the performance of their cameras and TCs. The small number who have used them say that they have found them relatively easy to use and easier to interpret than an image of the tapered lines of an ISO chart.As you will see from my responses to John Miles I’ve recently constructed an improved version of my test chart which has been developed to allow reasonably accurate assessment and comparison of cameras and lenses of all brands and include the influence of several of the factors you mention.See:http://www.dpreview.com/forums/post/52980923andhttp://www.dpreview.com/forums/post/52988984As you will see from the following PSE screen capture by using the Nikon 35 mm F/1.8 lens I bought in December with Live View and the D7100’s built-in flash I’ve been able to record an image of the latest version of my chart in which three single pixel lines separated by two single pixels spaces can be fairly easily recognised at several positions in the 1.0 pixel row.Crop from 100 ISO JPEG taken at F/5.6 and showing 1.0 to 1.4 pixels rows of black lines with 1.0 pixel lines and spaces indicated in red.I hope you will find the above image of some interest.JimmyJC -I presume your improved version is in monochrome with RGB and CMYK inking as per your post in the referenced threads?You are testing with perhaps the sharpest tool in the DX lens shed at this time (the FX Nikkors are better), but clearly the contrast has turned pretty much to mush at the 1.0 pixel level and the spaces between bars of the E are fairly evenly split between dark, light, and mid-grey. Note that you still have to climb back to about 1.3 before the pattern favors a 3 dark / 2 light sequence, and it's back to 1.4 before I'd say it is clearly that. And again, referencing photozone.de's tests of this lens on the D7000, we get 2737 LW/PH, and 2800 at f4, the peak for this lens. Conclusion: the 35mm looks to be a touch sharper than the 16-85, but tests at the same conditions would have to be run to be definitive.I can understand why users of your E chart would find it more satisfying than the tilted-edge Imatest pattern; your chart is designed for human viewing. Basically, what you've done is discretized the hyperbolic wedge test on the classic ISO 12233 test chart and turned the triple bars on the 1951 USAF High-Resolution test chart into a phalanx of shrinking letters. I do like it for its eye-chart friendliness, but computers like the tilted-edge pattern better.Thanks for taking the time to comment further on my tests and test charts.Yes, like the stepped Es chart in the referenced threads my new chart has been designed to be printed in RGBKCMY using my Canon iP4300 printer.With regard to your comment that the contrast has turned pretty much to mush at the 1.0 pixel level I was very pleased to read that you thought that the spaces between bars of the E are fairly evenly split between dark, light, and mid-grey.That has confirmed to me that the alignment of the camera to the chart was sufficiently accurate and that the nature of the resulting image is consistent with the behaviour I would expect from its sensor in relation to the design of the chart.You are correct in stating that my E charts are basically discretized versions of the classic ISO 12233 test chart in which I have used six colours in addition to black and used the letter E in place of the triple bars of the 1951 USAF High-Resolution chart or the five bars of the FUJIFILM test chart. However, as explained below, for the multiple stepped Es version there is a very important difference.When an individual looks at a letter E there is an interaction between the eyes and the brain in which the movement of the eyes adjusts the position of the image of the E on the retina until it is recognised by the brain.As a digital camera doesn’t have an equivalent control system when used to record an image of an E the position of the lines and the degree to which their edges overlap adjacent pixels will be entirely random.To compensate for that my charts have been designed so that the horizontal lines of each successive E in a group of eleven in each row is moved upwards by 0.1 of the thickness of the lines in the 1.0 pixel row.As a consequence the edges of the lines in at least one of the Es in each group of eleven in the 1.0 pixel row should overlap the edges of the pixels in the adjacent rows by less than +/- 5% of the height of a pixel and depending on their position within the group of eleven the edges of the lines in the other ten Es will overlap the adjacent rows by proportionately greater amounts.That in my opinion is the explanation for the range of shades of grey in the spaces between the lines of the Es in the 1.0 pixel row. A similar argument applies to the Es in all of the other rows.It is worth noting that it is only for the 1.0 and 2.0 pixel rows that it is possible for the edges of the lines in an E to be in exact register with the edges of a row of pixels.During the evolution and testing of my charts I realised that apart from making them recognisable as Es the vertical lines made no contribution to the assessment of the resolution of a camera and that due to the combined effects of overlapping adjacent pixels and the demozaicing of the Bayer matrix they could affect the pixels in the spaces between the horizontal lines.To cut an already long explanation short in the latest version of the chart the Es have been replaced by groups of three horizontal lines which are raised sequentially in 0.1 pixel steps like the stepped Es chart in the references in my previous post.In addition pixels to avoid an end effect problem observed with the three pixel length of the lines in the Es the length of the lines has been increased to ten pixels.I was very pleased to read that you like my chart for its eye-chart friendliness. While I appreciate that a tilted-edge pattern is more computer friendly I should point out that my charts have been developed to satisfy my own needs and to provide an inexpensive but very effective means whereby anyone with access to a decent colour printer and the patience to conduct the tests carefully can assess with acceptable accuracy the resolution of their cameras and lenses and the extent to which these are affected by a variety of factors.I hope that helps clarify the design and interpretation of my test charts.Jimmy

Cool, JC.  Have you published your "marching bar trio" test pattern update yet?  It might be fun to try it on my D90 and D7100 to see what happens.I appreciate the subtlety with which you have constructed your test target.  Your correspondents in the other threads were right to be impressed.Cheers-

rhlpetrus wrote:leerob wrote:according to Ken Rockwell, unless you have very good lenses. Does thids mean that lenses development lags behind sensor development. I'm talking about D7100 here.Actually any lens will perform better on the D7100 than on a lower pixed camera like the D7000. That's the case also when comparing the D800 x D610. You can see that in the measurement figures for D7000 x D7100 with same lens:http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7000__680http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7100__865So a 24 MP camera only gets 15 MP recorded?  Yes it's more than from the what, 16 MP d7000 but it isn't recording 24 MP of resolution if you believe DXO's numbers.

RedFox88 wrote:rhlpetrus wrote:leerob wrote:according to Ken Rockwell, unless you have very good lenses. Does thids mean that lenses development lags behind sensor development. I'm talking about D7100 here.Actually any lens will perform better on the D7100 than on a lower pixed camera like the D7000. That's the case also when comparing the D800 x D610. You can see that in the measurement figures for D7000 x D7100 with same lens:http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7000__680http://www.dxomark.com/Lenses/Nikon/Nikon-AF-S-DX-Micro-NIKKOR-40mm-F28G-mounted-on-Nikon-D7100__865So a 24 MP camera only gets 15 MP recorded? Yes it's more than from the what, 16 MP d7000 but it isn't recording 24 MP of resolution if you believe DXO's numbers.Believe. 24MP is the number of pixels in the image file, to be sure, but the Bayer CFA exacts a penalty in resolution.  That's why the Foveon sensor gets people so excited, even though its ISO limits are very low.

To leave all the math aside, a 24MP sensor will show wonderful real-world performance with a lens that provides it with sufficient resolution.  The practical problem is that Nikon no longer makesnewmodels of DX lenses with this quality unless its first number starts with 18, e.g., 18-140mm DX.  (I have done some comparisons on a few sites and find that it is about as sharp as my 16-85mm.  I have not had a 18-140mm in hand.)Of course, almost all new Nikkor FX lenses will provide the benefits to 24MP sensors on a DX camera.  They are just quite expensive for what you get.  (I think that includes the long ones, e.g., the new Nikkor 80-400mm model, even on sale, vs. the Tamron 150-600mm (yes, it's not yet out in Nikon mount, according to several recent, well regarded tests done on a Canon mount).)I believe that Nikon is exaggerating (in the marketing sense) the performance that almost all users of 24MP cameras will receive with the vast majority of Nikkor DX lenses they will use, with the exception of the 18-140mm kit lens, the occasional DX prime, and either of the DX micro lenses.  The full benefit of the 24MP sensor will come from the more expensive FX lenses, and (we can hope for more of the following) excellent third-party lenses.  Of course, relatively few DSLR users ever buy a lens other than the one kitted with their camera.IMO, more legitimate gripes on the current DX Nikons include buffer size, focus speed, and the like.

CrabbyGuy wrote:To leave all the math aside, a 24MP sensor will show wonderful real-world performance with a lens that provides it with sufficient resolution. The practical problem is that Nikon no longer makesnewmodels of DX lenses with this quality unless its first number starts with 18, e.g., 18-140mm DX. (I have done some comparisons on a few sites and find that it is about as sharp as my 16-85mm. I have not had a 18-140mm in hand.)The 40mm macro seems to meet that quality need/desire. Not sure what the future holds though. I do certainly prefers FX lenses on my DX bodies but not because of pure IQ. I see FX lenses as really FX/DX and DX lenses as designed to be more economical.Of course, almost all new Nikkor FX lenses will provide the benefits to 24MP sensors on a DX camera. They are just quite expensive for what you get. (I think that includes the long ones, e.g., the new Nikkor 80-400mm model, even on sale, vs. the Tamron 150-600mm (yes, it's not yet out in Nikon mount, according to several recent, well regarded tests done on a Canon mount).)I believe that Nikon is exaggerating (in the marketing sense) the performance that almost all users of 24MP cameras will receive with the vast majority of Nikkor DX lenses they will use, with the exception of the 18-140mm kit lens, the occasional DX prime, and either of the DX micro lenses. The full benefit of the 24MP sensor will come from the more expensive FX lenses, and (we can hope for more of the following) excellent third-party lenses. Of course, relatively few DSLR users ever buy a lens other than the one kitted with their camera.I have not scene any Nikon lens that doesn't result in a better overall image on the new 24mp sensor than the older gen (no absolutes). Maybe not full potential...but certainly improved. No disrespect, but I just don't see a current Nikon DX lens the doesn't improve, vs 16mp,  on the current gen 24mp....when printed to same size/framing. Even the older 18-105 looks better on 24mp.

leerob wrote:according to Ken Rockwell, unless you have very good lenses. Does thids mean that lenses development lags behind sensor development. I'm talking about D7100 here.First off, Ken Rockwell is the Michael Jordan of bad advice. Never take his word as gospel.As someone who has shot with Nikon D70, D90 and D3200, the D3200 produces the sharpest and best images out of all three. You have to actually know what you're doing though...And finally, you ALWAYS need good lenses to produce good images. This is the same across the board--DX, FX, Canon, Nikon, Fuji. Your lens is more important than the camera body.

I agree that lens is very important. Wish I could justify buying Leica.