thebestcpu
Fan of Printing
- Joined
- Dec 8, 2024
- Messages
- 112
- Reaction score
- 64
- Points
- 60
- Printer Model
- Epson SC P900
My goal is to establish for me my go-to PPI in Photoshop. So, the experiments were for me, yet I am sharing my results as there may be some interest. As always, I am open to feedback, constructive comments, and corrections to what I have shared below.
While building a set of test images to evaluate the Epson P900, I observed a subtle but significant issue in the downsampling process that isn’t often discussed: detail loss when downsampling from high-resolution sources.
As a typical example, if you're working with a very high-resolution image and intend to preserve detail at the 300 or 360 PPI level, you might naturally downsample directly to that target resolution. However, this approach may discard important details based on the Nyquist criterion.
The Nyquist theorem states that the sampling rate must be at least twice the maximum frequency of the signal to preserve signal integrity in any sampled system. For instance, CDs are sampled at 44,100 Hz in digital audio to capture audio up to approximately 20,000 Hz.
The same principle applies to images. High-frequency detail in photographs—fine texture, tight patterns, etc.—requires equivalent sampling logic. To preserve image detail corresponding to 300 or 360 PPI, you should downsample to 600 or 720 PPI, respectively.
Visual Demonstration: Checkerboard Pattern Downsampling
To illustrate this, I created synthetic checkerboard patterns simulating high-frequency detail, ranging from 2880 PPI down to 90 PPI in even steps. Checkerboard patterns are incomplete for resolution testing yet are very good for specific tests:
Next, I resampled the image to three target resolutions—1440, 720, and 360 PPI—to examine how different frequencies reproduce. Each test box includes a zoomed inset for easier viewing, with the magnification labeled for reference.
1440 PPI Downsample
At 1440 PPI, early signs of the checkerboard pattern emerge, albeit with low contrast and clarity. At 960 PPI, checkerboards reappear but are distorted—different from the original pattern. At 720 PPI and below, detail reproduction improves significantly, with 720 PPI offering an accurate representation of the original checkerboard—consistent with Nyquist's rule (1/2 of 1440 PPI).
I've marked the results accordingly:
While checkerboards exaggerate the effect, photographic images may mask this degradation with more random pixel structures. That said, if you’ve invested in high-resolution cameras, lenses, or scanners to capture fine detail, it’s worth ensuring your downsampling method preserves that fidelity.
This pattern holds in further tests: 720 and 360 PPI resamples only reproduce details reliably when targeting half the PPI value of the original content.
Below is a zoomed view of the 360 PPI downsample at the 360 PPI checkerboard frequency. Despite being numerically matched, the pattern is neither clean nor well-resolved over a larger area—supporting the need to downsample at twice the detail frequency you wish to preserve. The magnified inset just happened to miss some of the other anomalies.
Takeaway
Whether this level of detail loss matters depends on your needs and standards. However, many users may inadvertently discard quality due to insufficient downsampling resolution. This test shows subtle detail loss can occur before the image reaches the printer.
In a future post, I’ll take these digital samples to print and compare the differences in real-world output.
The key takeaway for me is that my go-to Photoshop PPI for the Epson p900 is leaning to be 720 PPI as I want to preserve the detail I can at 360 PPI, which is about the resolution of the human eye.
Always open for feedback.
John Wheeler
While building a set of test images to evaluate the Epson P900, I observed a subtle but significant issue in the downsampling process that isn’t often discussed: detail loss when downsampling from high-resolution sources.
As a typical example, if you're working with a very high-resolution image and intend to preserve detail at the 300 or 360 PPI level, you might naturally downsample directly to that target resolution. However, this approach may discard important details based on the Nyquist criterion.
The Nyquist theorem states that the sampling rate must be at least twice the maximum frequency of the signal to preserve signal integrity in any sampled system. For instance, CDs are sampled at 44,100 Hz in digital audio to capture audio up to approximately 20,000 Hz.
The same principle applies to images. High-frequency detail in photographs—fine texture, tight patterns, etc.—requires equivalent sampling logic. To preserve image detail corresponding to 300 or 360 PPI, you should downsample to 600 or 720 PPI, respectively.
To illustrate this, I created synthetic checkerboard patterns simulating high-frequency detail, ranging from 2880 PPI down to 90 PPI in even steps. Checkerboard patterns are incomplete for resolution testing yet are very good for specific tests:
Next, I resampled the image to three target resolutions—1440, 720, and 360 PPI—to examine how different frequencies reproduce. Each test box includes a zoomed inset for easier viewing, with the magnification labeled for reference.
1440 PPI Downsample
At 1440 PPI, early signs of the checkerboard pattern emerge, albeit with low contrast and clarity. At 960 PPI, checkerboards reappear but are distorted—different from the original pattern. At 720 PPI and below, detail reproduction improves significantly, with 720 PPI offering an accurate representation of the original checkerboard—consistent with Nyquist's rule (1/2 of 1440 PPI).
I've marked the results accordingly:
- Black box: Unusable/distorted
- Orange box: Some recognizable detail, but degraded
- Yellow box: Caution—partial recovery, still distorted
- No box: Good reproduction
While checkerboards exaggerate the effect, photographic images may mask this degradation with more random pixel structures. That said, if you’ve invested in high-resolution cameras, lenses, or scanners to capture fine detail, it’s worth ensuring your downsampling method preserves that fidelity.
This pattern holds in further tests: 720 and 360 PPI resamples only reproduce details reliably when targeting half the PPI value of the original content.
Below is a zoomed view of the 360 PPI downsample at the 360 PPI checkerboard frequency. Despite being numerically matched, the pattern is neither clean nor well-resolved over a larger area—supporting the need to downsample at twice the detail frequency you wish to preserve. The magnified inset just happened to miss some of the other anomalies.
Whether this level of detail loss matters depends on your needs and standards. However, many users may inadvertently discard quality due to insufficient downsampling resolution. This test shows subtle detail loss can occur before the image reaches the printer.
In a future post, I’ll take these digital samples to print and compare the differences in real-world output.
The key takeaway for me is that my go-to Photoshop PPI for the Epson p900 is leaning to be 720 PPI as I want to preserve the detail I can at 360 PPI, which is about the resolution of the human eye.
Always open for feedback.
John Wheeler
