Originally Posted by Jones_Rush
What I meant is, if you want to compare 720 to 1080, you should ideally use the same displaying device (which is capable of producing all resolutions), and only change the resolution. All the variables apart from resolution should be kept constant. So, if you want to find out the influence of DPI on the quality of the image, but do this while comparing an LCD monitor to a projector, that's bad practice.
I think the way I did it in post #21 is better.
The problem with using a CRT and changing resolutions is that a CRT isn't a fixed pixel device. Whereas a 720 image could use more than one dot per pixel, a 1080 image would not.
The method I used (explained at the bottom of the post) will work with any monitor - CRT or LCD. I may not have been clear enough in post 21, so I'll try to explain it a little better here. In fact, I'm going to try to explain it so clearly that anybody with enough patience to read through all this can do it. I have also explained in detail the process and reasoning behind this method, again at the bottom of the post. This way if there is any debate as to the findings of this particular method, the process and reasoning is spelled out for critique and debate. For this example I will be referring to the Photopshop image:hdcomp.psd (18.5MB)
right-click, and select "save as". Save it wherever you want, but remember where you put it.
I will explain the reasoning behind this method at the end of this post.
First, you need to ensure your monitor is set at a resolution which will allow you to see all the pixels. And so, look at this image:
Here is a larger version of the image.
If you cannot distinguish all these details in the smaller image, your resolution is set too high, and must be lowered. On an LCD monitor set to it's native resolution, the first image should be crystal clear. If the dots seem to blend into one, then your CRT resolution is set too high, and you aren't seeing all the information. If this is happening with an LCD monitor, you are not set to your monitor's native resolution.
Okay, on to the hdcomp.psd file.
In order to view this file correctly, you need software which will not only display .psd images, but which will also allow you to turn the individual layers on and off.
Now that your monitor has been set so that you can distinguish a space between the two dots, open the hdcomp.psd file.1.
Determine how far you should be from your monitor. Since the image is spilling off the sides of the screen, you need to sit a little farther than 1.5X from your screen. Use this table (explanation follows)
Resolution / / Monitor Widths / / equivalent 16x9 Screen Widths
800x600 / / 2.4 / / 1
800x600 / / 3.6 / / 1.5
800x600 / / 4.8 / / 2
1024x768 / / 1.9 / / 1
1024x768 / / 2.8 / / 1.5
1024x768 / / 3.8 / / 2
1280x720/960/1024 / / 1.5 / / 1
1280x720/960/1024 / / 2.25 / / 1.5
1280x720/960/1024 / / 3 / / 2
1600x1200 / / 1.2 / / 1
1600x1200 / / 1.8 / / 1.5
1600x1200 / / 2.4 / / 2
1920x1080 / / Duh
2048x1536 / / .93 / / 1
2048x1536 / / 1.4 / / 1.5
2048x1536 / / 1.86 / / 2
How to use this chart: First, determine the resolution your monitor is set at. For those who are unsure, right-click on your desktop, select "properties", and select the "Display Settings" tab. The resolution should be listed in "Screen Resolution". Don't change it, just remember where it's at.
Now, find your screen resolution in the table above. Now, choose the screen width closest to wherevver you sit (or would sit) from your home theater screen. This is the number in the 3rd column. You can pick either 1 (screen width), 1.5 (screen widths) or 2 (screen widths).
Once you have selected this, look at the number in the second column. This is how far you should sit from your monitor, in screen widths. Use a tape measure to determine the width of your monitor, and multiply by the screen widths to determine how far your eye should be from the monitor.
Screen resolution: 1280x960
Let's say I want to simulate sitting 1.5 screen widths from my home theater screen. So, here is how I determine how far to sit from my monitor when viewing the hdcomp.psd file.
I determined that I want to see what 1.5 screen widths look like. So, in the 8th row of the chart, I find this:
1280x720/960/1024 / / 2.25 / / 1.5
The 2.25 in the second column tells me I need to sit 2.25 screen widths from my monitor in order to simulate a viewing distance of 1.5 screen widths from a 16:9 home theater screen. (remember, the hdcomp.psd file scaled to 100% is spilling off the sides of my screen)
Now, I take a tape measure and measure the width of my monitor.
14 3/8 inches. 3/8 of an inch equals .375 inches. (3 divided by 8)
So, my monitor is 14.375 inches wide, and I need to sit 2.25 screen widths to simulate sitting 1.5 screen widths from a 16:9 image.
14.375 x 2.25 = 32.34375 inches.
For simplicity sake, we'll round down to a nice even 32 inches.
So, I sit so that my eyes are 32 inches from the front of the screen. This will simulate sitting 1.5 screen widths away from a 16:9 image.2.
Seat yourself at the distance you determined in step 1. Open the hdcomp.psd file in Photoshop, or some variant thereof (CS, LE, 3.0, 4.0, etc). If you do not have Photoshop, skip to the ALTERNATE METHOD below.3. CRITICAL - set the image scaling to 100%.
Chances are the edges of the image will be spilling off the screen. This is fine. We don't need to see the whole image all at once in order to do the test.4.
There should be three layers. One labeled "854x480", one labeled "1280x720", and one labeled "background". The background image is 1920x1080. The layer you are seeing when the picture is first opened will be the 854x480 layer.5.
Hide the 854x480 layer. Now you are viewing the 1280x720 layer. Turn the 854x480 layer back on. Toggle it several times, and you should see the clear difference between standard definition and 720x1280.6.
Hide the 854x480 layer. Now hide the 1280x720 layer. You are now viewing the full resolution 1920x1080 layer. Flip the 1280x720 layer on and off several times. Depending on how far you are sitting, you should notice a difference in sharpness. For me, the difference is clear at 1 screen width. The difference is minor at 1.5 screen widths. I haven't tried it at 2 screen widths, but I doubt I would see a difference.ALTERNATE METHOD2.
Click on the below link. A new window will open with the picture.hdcomp.jpg
If your resolution is set lower than 1920x1080, the entire picture should NOT fit on the screen. If the image has been re-sized, move your mouse to the lower right corner of the image. A box should appear. Click this box, and the image will be shown at 100% size. If it now spills off your screen and all over your computer desk, this is a good thing.
If the above doesn't work to get the image to display at full size, in Internet Explorer, click on Tools-->Internet Options --> Advanced. Scroll down to the "Multimedia" section. The first option listed should be "Enable Automatic Image Re-Sizing". Uncheck this box and re-load the image. It should now spill off the screen.
Simply scroll up and down to view the different images. They are labled for your convenience.Image Explanation
hdcomp.psd was created using a 2048x1536 image taken with a digital camera (Olymps C3030). This image was then cropped to a resolution of 1920x1080. The resulting image was then saved.
Next, I re-sized the image to 1280x720. After doing this I again re-sized it to 1920x1080 without filtering. This resulted in a slightly pixelated 1920x1080 image. This simulates what a 720x1280 image would look like compared to a 1920x1080 image of the same physical size. This image was saved.
Next, I re-sized the original 1920x1080 image to 854x480, and again re-sized it back up to 1920x1080, again without filtering. The resulting image is even more clearly pixelated. This image was saved.
Finally, I added all three images to one photoshop file so that there are three layers. The top layer being 854x480 scaled up to 1920x1080 without filtering, the second layer being 720x1280 again scaled up to 1920x1080 without filtering, and the third "background" layer being the original 1920x1080 image.
QUESTION 1: Why resize a low resolution image to a high resolution image WITHOUT filtering?
What this experiment is supposed to simulate is three digital projectors which are exactly the same with the exception of resolution. One projector has a native resolution of 854x480. One has a native resolution of 1280x720. One has a native resolution of 1920x1080. If you use the 480p projector on a 100" screen, it will be pixelated. If you use the 1080p projector on the same screen, it will appear smoother.
And so, the 854x480 image was re-sized to 1080x1920 without any sort of filtering in order to simulate what the pixelated image would look like. Likewise with the 1280x720 image.
QUESTION 2: Why not just use three images - 854x480, 1280x720, and 1920x1080, and display them full screen using different monitor resolutions?
The first reason is because a direct comparison using layers in a Photoshop file allows direct instantaneous comparison of each resolution by simply turning a layer on and off. That way we don't have to try to remember what the image looks like while we close the image, go into display properties, set the new resolution, apply the new resolution, and open the new file.
CRT: The reson not to do this with a CRT monitor is because most monitors do not really show all the detail possible at their highest resolution. The image of the two dots near the top of this post demonstrates that. To do this experiment accurately, the monitor MUST be set at a resolution which will resolve all the pixels. If all the pixels are not resolvable, then image detail is lost, and a 1920x1080 image may very well NOT look any better than a 1280x720 image.
This is why in my method the monitor is set at it's optimal resolution, and the images are viewed, even though the entire image is not shown on the screen. This isn't about viewing the entire image, it's about viewing enough of the image to notice detail differences.
LCD: You don't change resolutions with LCD quite simply because they are fixed pixel devices, meaning that no matter what resolution you set your desktop at, you are always seeing the same number of pixels. As such, any resolution higher than your monitor's native resolution will not look any better than it's native resolution.SCREEN WIDTHS
To determine the values in the table used to determine the distance to be used when viewing the image, some simple mathematics were used. For example:
1024x768 / / 2.8 / / 1.5
The screen resolution and seating distance of 1.5 are given. What needs to be done is the monitor widths (2.8) needs to be extrapolated. In order to do this, you need to determine the ratio between the horizontal image resolution and the monitor resolution. Therefore:
1920 divided by 1024 equals 1.875. This is how wide the overall image is in relation to your monitor. So, if our monitor is 14 inches wide, the actual image would require 26.25 inches (14 times 1.875) of monitor to display as the actual image is 26.25 inches wide.
Now that we know that at 1024x768 the resulting image is 1.875 screen widths, we multiply this number by the desired seating distance. In this case, 1.5 screen widths.
1.875 times 1.5 equals 2.8125, rounded to 2.8 for simplicity sake.
So, if we take our monitor's measured width of 14 inches and multiply by 2.8, we get 39.2 inches, which is the distance we need to sit in order to simulate 1.5 screen widths of the 16:9 image.
To check this, we know that the entire image is 1.875 screen widths wide, or 26.25 inches. If we multiply 26.25 by our desired seating distance of 1.5 screen widths, we get 39.375 inches. The first number we determined for seating distance was 39.2. The difference is because we rounded 2.8125 to 2.8. If we eliminate the rounding and multiply 14 by 2.8125, we indeed get 39.375.LIMITATIONS
Although I try as accurately as possible to simulate three separate projectors, the test is not perfect. 1 720p pixel does not translate to 4 1080p pixels. Nor does 1 480p pixel translate evenly to 4 720p pixels. Mathematically, 1 720p pixel would be the same size as 2.25 1080p pixels. And 1 480p pixel would be the same size as 2.25 720p pixels, or 5.06 1080p pixels. Unfortunately, digital pictures don't work that way. The end result is that this experiment will not show as much of a detail difference as would be seen in the real world. Ideally we would have a side by side by side comparison of a 480p, 720p, and 1080p projector. But since most of us can't afford to do this, and since projector retailers aren't in the rental business, we have to find alternate means.
I believe the method I listed above is the best means for everybody on this forum to see for themselves if a 1920x1080 projector will offer any benefit for them. (short of buying three separate projectors) It allows direct comparison of three images of the same physical size and resolution, yet three separate levels of detail roughly equivalent to 480p, 720p, and 1080p.