I had recently noticed that the actual resolution of the consumer HDV cameras was around 600 x 600, even though they are recording to 1080 x 1440 pixels (example Canon HV30 http://www.camcorderinfo.com/content/Canon-HV30-Camcorder-Review-34401/Performance.htm). I assumed that, as the lens and sensors got better, the actual resolution went way up. Looking at a comparison of prosumer cameras, the measured resolution from the test charts showed only slightly better resolution on most the cameras here (http://www.dv.com/features/features_item.php?articleId=192501232). I hope Im missing something here. Can someone please explain?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Actual resolution vs. pixels on prosumer cameras?
- Thread starter Chris7
- Start date
- Sort by reaction score
Thanks. Perhaps I was not being clear.
According to camcorderreview, the HV30, which records to 1440 x 1080 (about 1.5 megapixels of data) was able to produce a horizontal resolution of 625 line widths per picture height (lw/ph) and a vertical resolution of 575 lw/ph.
In the Digital Video Texas Shootout, he Canon XL H1, which also records to 1440 x 1080 (in 60i) appears that it is only able to pick up about 800 lines horizontally and over 700 vertically (in 60i), when shooting test charts in standard video gamma and standard color matrices.
I guess I had assumed a bigger jump in actual resolution with a camera that is four pounds heavier and several thousand dollars more expensive.
1) I should have asked, what are you gaining when going from consumer to prosumer cameras (mainly just better color?).
2) Additionally, say I bought a Red Scarlet 1/3 down the road (which shoots at up to an estimated 3,000 x 2300, about seven megapixels of data) What would you speculate its actual resolution to be? How about once I transferred it to 1080 x 1920?
Many thanks.
And that's the reason everyone went ga-ga over the HV20. At the time, and this still may be true, it offered the best image from a consumer camera by far. Many people thought that Canon had 'goofed' and released a better quality camera than they meant too but obviously that wasn't the case since the HV30 didn't get dumbed down. W/that being said you have to take different tests performed by different people w/a grain of salt as the testing environments are different and there is some amount of subjectivity when evaluating test charts. Also, shooting test charts isn't the same as shooting 'real world' footage and that's why in the DV.com article they have examples of both.
Some things you get in the XH-L1 is better low light performance, better dynamic range, a better lens, interchangeable lenses, professional hook-ups (HD-SDI, TC, genlock, XLR, etc.,), controls to dial in specific 'looks' in camera, etc,. There might also be differences in image quality if the XH-L1 is using better analog-to-digital converters and/or better on-board encoding hardware.
The 2/3" Scarlet is rumored to have an effective resolution of about 2K (depending on the quality of the lens on the camera). 2K is a bit bigger than 1080 so deriving a 1080 image from Scarlet should result in higher image quality than shooting w/a native 1080 camera because of the inherent loss of image quality that happens during the recording process (all other things being equal of course).
Lethal
Red Scarlet
OK, that makes sense, thanks.
1) To clarify, is the rumored effective resolution of the Scarlet 2/3" (thanks for pardoning my typo, BTW), to be about 2K horizontally (depending on lens)? What is the rumored vertical resolution, or does the 2K refer to both vertical and horizontal effective resolution? (Now I see what the hype is about -- this is over two and a half times the effective resolution as the prosumer cameras).
2) Assuming that the effective resolution is 2K horizontally when shooting 3K, would most of this resolution be preserved when transferring the data to 1920 x 1080? If no, what would you speculate would be the effective res after conversion?
3) How about if I shot in 2K (which the Red One is capable of shooting in, and I assume the Scarlet will be also). Would I get the full 2K effective resolution also? If not, what effective resolution would you speculate if shooting in 2K?
Thanks in advance.
OK, that makes sense, thanks.
1) To clarify, is the rumored effective resolution of the Scarlet 2/3" (thanks for pardoning my typo, BTW), to be about 2K horizontally (depending on lens)? What is the rumored vertical resolution, or does the 2K refer to both vertical and horizontal effective resolution? (Now I see what the hype is about -- this is over two and a half times the effective resolution as the prosumer cameras).
2) Assuming that the effective resolution is 2K horizontally when shooting 3K, would most of this resolution be preserved when transferring the data to 1920 x 1080? If no, what would you speculate would be the effective res after conversion?
3) How about if I shot in 2K (which the Red One is capable of shooting in, and I assume the Scarlet will be also). Would I get the full 2K effective resolution also? If not, what effective resolution would you speculate if shooting in 2K?
Thanks in advance.
1) I'm assuming the 2K people are talking about in this instance is 2048x1080. So it's an imperceptible amount more than 1080p. But if you have the capacity to process and output it as such it can be full RGB, making it potentially a nice amount sharper and richer than a 1080p Blu-ray.
2) 1920x1080 of it would be.
3) It shoots windowed 2K, so no. You'd lose a third of the 2K (from 3K bayer) resolution. You do the math(s).
P.S. The "hype" is just that. Hype. It doesn't yet exist as a saleable product. Things could change. Count on it.
2) 1920x1080 of it would be.
3) It shoots windowed 2K, so no. You'd lose a third of the 2K (from 3K bayer) resolution. You do the math(s).
P.S. The "hype" is just that. Hype. It doesn't yet exist as a saleable product. Things could change. Count on it.
The RED ONE is capable of 4k and 2k while the Scarlet can/will shoot up to 3k. Of course, since the Scarlet is still vaporware at this point, the specs are subject to change. For a comparison chart of resolutions, see: http://www.red.com/cameras/technology/
For a while, the RED website mentioned another camera in the works capable of 5k, but I can no longer find that info on their website.
Keep in mind that several Asian television manufacturers are already in product development of UHD TVs (Ultra High Definition - also known as "Super Hi-Vision") and NHK (Japan's Broadcasting Corporation) has already run UHDTV test broadcasts via satellite at 7680x4320 pixel resolution with progressive scanning at 60 frames per second. http://www.eetimes.com/showArticle.jhtml?articleID=188500682
-DH
1. From what I've read the Red One hits about 1800 l/ph on a test chart (about 3.2K effective resolution). So a safe ball park number for the 2/3" Scarlet could be about 1200 l/ph. But that's just speculation because Scarlet doesn't exist currently.
2. That depends on the quality of the software and/or hardware doing the conversion.
-DH,
On the 13th Red released some new info on Scarlet and Epic.
Lethal
2. That depends on the quality of the software and/or hardware doing the conversion.
-DH,
On the 13th Red released some new info on Scarlet and Epic.
Lethal
Sorry, I'm not familiar with "l/ph" or how this is translated into 3.2K (horizontal/vertical?) effective resolution. Could you please explain?
I'll attempt to butcher some information. Others .. please correct me as needed.
Resolution in "lines" is about 1/2 the pixel count.
For example a computer screen is 1280x800. My laptop. If I drew 1280 vertical black lines on the screen, all you would see is black. No lines. If I drew 1 black then 1 white vertical lines etc across the screen. There would be 640 vertical white lines and 640 vertical black lines. Hence 1/2 screen pixel count.
I don't think the resolution can be greater than 1/2 the pixel count on a particular axis. I could be wrong here. There's a thing called MTF or Modulation Transfer Function. There is info on the web.
So now apply this to a camera imaging a test pattern. As the line resolution gets finer (more lines per ...) the black and white lines will go from distinct black and white lines to gray.
I hope this has not damaged anyone.
Resolution in "lines" is about 1/2 the pixel count.
For example a computer screen is 1280x800. My laptop. If I drew 1280 vertical black lines on the screen, all you would see is black. No lines. If I drew 1 black then 1 white vertical lines etc across the screen. There would be 640 vertical white lines and 640 vertical black lines. Hence 1/2 screen pixel count.
I don't think the resolution can be greater than 1/2 the pixel count on a particular axis. I could be wrong here. There's a thing called MTF or Modulation Transfer Function. There is info on the web.
So now apply this to a camera imaging a test pattern. As the line resolution gets finer (more lines per ...) the black and white lines will go from distinct black and white lines to gray.
I hope this has not damaged anyone.
According to camcorderreview, the HV30, which records to 1440 x 1080 (about 1.5 megapixels of data) “was able to produce a horizontal resolution of 625 line widths per picture height (lw/ph) and a vertical resolution of 575 lw/ph.”
Does the horizontal 625 line widths per picture height of the Canon HV30 translate into higher effective resolution?
Does the horizontal 625 line widths per picture height of the Canon HV30 translate into higher effective resolution?
Unfortunately I don't know enough about measuring how much detail a camera can resolve to be able to really clarify things (I'm more of a post guy and only know enough about production to be dangerous). Part of the problem is some people talk about how much resolution can be resolved in traditional, analog TV terms (lines, or line pairs, per picture height), some people talk about it in traditional film terms (lines per picture width or lines per millimeter) and some people talk about it in digital cinema terms (2k, 4k, etc,.). Which of course can lead to a lot of confusion. It's kinda like people talking about temperature in Fahrenheit, Celsius, and Kelvin and assuming everyone knows how to readily convert between all three. The waters further get muddied because the lens, sensor, codec, test conditions and in-camera settings all impact the quality of the image. Also, like I said before, test results can be subjective.
If you feel like some nice, heavy reading here are some links you might enjoy.
Three three-letter cameras: EX1, F23, RED
More RED Res Testing: The Mystery(ium) Resolved
Wikipedia on Digital Cinema
RedUser.net: True Resolution of Mysterium difficult to determine.
RedUser.net: Red One Measured Resolution
Flynnstone,
AFAIK, what you are describing is measuring using 'line pairs' as opposed to just using 'lines'. For example, 1200 lines is the same as 600 line pairs. Why the two different terms? I have no idea. You are correct though that a sensor cannot resolve an image greater in detail then itself. By that I mean, a 1920x1080 sensor cannot resolve anything beyond 1920x1080 (and in the real world it can't even resolve 1920x1080).
Lethal
If you feel like some nice, heavy reading here are some links you might enjoy.
Three three-letter cameras: EX1, F23, RED
More RED Res Testing: The Mystery(ium) Resolved
Wikipedia on Digital Cinema
RedUser.net: True Resolution of Mysterium difficult to determine.
RedUser.net: Red One Measured Resolution
Flynnstone,
AFAIK, what you are describing is measuring using 'line pairs' as opposed to just using 'lines'. For example, 1200 lines is the same as 600 line pairs. Why the two different terms? I have no idea. You are correct though that a sensor cannot resolve an image greater in detail then itself. By that I mean, a 1920x1080 sensor cannot resolve anything beyond 1920x1080 (and in the real world it can't even resolve 1920x1080).
A higher effective resolution as compared to what? Effective resolution is like your paycheck after taxes. Sensor resolution is like your paycheck before taxes.
Lethal
Thanks for the links. I read a good part of all of them. Anyone please tell me if Ive got this right:
A RED ONE gets 1800 lw/ph, which translates into 3.2K cine number or effective resolution. From first link, I inferred that lw/ph is multiplied by two to get the effective resolution (quote below), so I still dont understand why 1800 lw/ph does not translate to 3.6K, but instead gets 3.2K (which is 1800 multiplied by 1.77). So if the Canon HV30 gets a horizontal resolution of 625 line widths per picture height (lw/ph), this should translate into 1250 (1.25K) effective resolution (or 1100 if I multiply it by 1.77). This is excellent for a camera whos sensor only picks up 1440 lines horizontally.
When we talk about TV, we talk about TVl/ph: TV lines per picture height. When we talk about digital cinema, we talk about a 2k image or a 4k image as images with that many pixels across. To convert TVl/ph to horizontal pixels, we multiply by the aspect ratio of the picture. As we shot the RED at a 2:1 aspect ratio, we can read the TV number off the resolution figures on the chart and multiply by two to get the cine number. http://provideocoalition.com/index.php/awilt/story/three_three_letter_cameras_ex1_f23_red/P2/
(I assume that lines widths per picture height (l/ph) and TV lines per picture height (TVI/ph) are the same thing.)
A RED ONE gets 1800 lw/ph, which translates into 3.2K cine number or effective resolution. From first link, I inferred that lw/ph is multiplied by two to get the effective resolution (quote below), so I still dont understand why 1800 lw/ph does not translate to 3.6K, but instead gets 3.2K (which is 1800 multiplied by 1.77). So if the Canon HV30 gets a horizontal resolution of 625 line widths per picture height (lw/ph), this should translate into 1250 (1.25K) effective resolution (or 1100 if I multiply it by 1.77). This is excellent for a camera whos sensor only picks up 1440 lines horizontally.
When we talk about TV, we talk about TVl/ph: TV lines per picture height. When we talk about digital cinema, we talk about a 2k image or a 4k image as images with that many pixels across. To convert TVl/ph to horizontal pixels, we multiply by the aspect ratio of the picture. As we shot the RED at a 2:1 aspect ratio, we can read the TV number off the resolution figures on the chart and multiply by two to get the cine number. http://provideocoalition.com/index.php/awilt/story/three_three_letter_cameras_ex1_f23_red/P2/
(I assume that lines widths per picture height (l/ph) and TV lines per picture height (TVI/ph) are the same thing.)
1800 lw/ph is the effective resolution but just stated in a different way.
Double check what you quoted and you'll se that you multiple by the aspect ratio, which in that specific test by Adam Wilt happened to be 2:1. During this test Adam's test charts were limited to 1200 l/ph so he had to settle for an incomplete conclusion of saying Red One could resolve images at 2.4K+ because his charts didn't go any higher. Again, this goes back to paying close attention to the testing methodology and the dangers of mixing and matching results from different tests. The other tests that showed Red One resolving 1800l/ph were done w/a 16:9 aspect ratio (1.78) so that's why we get 3.2K in 'cine terms' and not 3.6K. Just to add another wrinkle to keep an eye out for here is post #4 from second RedUser link in my previous post:
Andrew said:
The HV30's sensor is 1920x1080, it's just the HDV format that is 1440x1080.
I do think it's just different nomenclature so I assume that is a safe assumption.
Lethal
Thanks again, Lethal, this really helps. I do not understand the utility of using "effective resolution" or "cine number," – it would be most easy for me to think of resolution in simple terms of double the lw/ph, even when talking about a picture that is not 2:1. (I’m sure there is a reason “cine number” and “effective resolution” are used, I just don’t get it).
I’m starting to see what you mean…
At this point, my biggest questions for anyone here are:
1. How do I interpret the test chart data from camcorderenfo.com?
2. How do I interpret the test chart measurements from DV.com (Texas Shootout)?
3. How do I compare these measurements to find the difference in resolution between a Canon HV30 and a XL H1?
Anyone please tell me if I've got this right:
The RED ONE is able to read 1800 black lines horizontally on a chart, which one could think of as 3600 "lines" of "actual resolution" (my term – I'm just doubling the number by adding the black and white lines here).
According to camcorderinfo.com, the Canon HV30 can show 625 lines horizontally and 575 lines vertically on a test chart. This translates into "cine number" or "effective resolution" of about 1100, and a horizontal "actual resolution" (again, my term, meaning lw/ph x 2) of 1250, and a vertical "actual resolution" of 1050. (I'm starting to doubt the accuracy of this, as the sensor can only pick up 1080 pixels vertically). http://www.camcorderinfo.com/content/Canon-HV30-Camcorder-Review-34401/Performance.htm#
BUT, a Canon XL H1 gets a horizontal res of 800 and a vertical res of 700+, according to the Texas Shootout. http://www.dv.com/features/features_item.php?articleId=192501232 I have no idea what to make of this figure. Is it in lw/ph? If I doubled the horizontal 800, I would get 1600. Great. But if I double the vertical 700+, I get 1400+, which is more than the sensor even has pixels for.
Can someone shed some light on this?
Think of effective resolution as temperature and l/ph and 'cine number' as Fahrenheit and Celsius. They are just two different ways to measure the same thing. For example, 1800 l/ph equals 3.2K in the same way that 32 degrees Fahrenheit equals 0 degrees Celsius.
Don't double the lines. 1800 lines per picture height (l/ph) counts each black and white line separately. If you count the lines together you would call it line pairs per picture height (lp/ph). So Red One has an effective resolution of 1800 l/ph aka 900 lp/ph.
I think that review should say "l/ph" (lines per picture height) not "lw/ph" (line widths per picture height). lw/ph seems to be a measurement used for still cameras and the test chart they are using says the unit of measure is lines per picture height (aka TV lines per picture height). You can't read it in their review, but if you click on an image in the DV.com Four Way Shoot Out you can read what is written on the chart.
Lethal
Thanks, Lethal. I think this clears it up, but someone correct me if Im wrong: The Canon HV30 can resolve about 625 horizontal lines of information, the XL H1 gets 800, and a RED ONE gets 1800. Cine numbers and effective resolution just confuse things for me, but anyway, these three cameras can be compared using the these three numbers. So even the best cameras are only able to resolve less than half of their horizontal pixels.