You swap a 2 MP, 1/4-inch mono camera for a 5 MP part of the same optical format, keep the lens and the f-number, and the image turns gray: less contrast, less bite in the fine detail. The new camera is not defective. Its pixels are smaller, so they sample spatial frequencies where the same lens passes far less modulation, and the 10- or 12-bit output may also be reaching your software with the wrong bit mapping. Below are the numbers that decide how much contrast a lens can hand a given pixel pitch, the settings to rule out first, what opening the aperture buys and costs, when only a different lens will do, and the decision for each case.
Why does a higher-resolution camera on the same lens look grayer?
Because the pixels are smaller, the sensor samples spatial frequencies where your lens passes far less contrast: at f/4 and 550 nm, diffraction leaves about 39% modulation at the 227 lp/mm the old 2.2 µm pixel needs, but only 20% at the 313 lp/mm the new 1.6 µm pitch needs.
The lens did not change; the question you ask of it did. Contrast at the sensor is set by the lens MTF, the fraction of a bright-dark swing at each spatial frequency that survives the optics. MTF falls steadily with frequency and reaches zero at the diffraction cutoff, so the finer site reads the same curve 38% further out, on the steep part, where a perfect lens delivers about half the contrast the larger one enjoyed.
Smaller full wells add a second flattening: each site holds and collects less charge, so shot noise takes a bigger share of the histogram, and display gamma exaggerates the flatness into a washed-out look. Choosing pixel size a lens can actually feed, from the other side of the same trade, is covered in small-pixel camera selection and in the Nyquist limit pairing of sensor and lens.
Which camera settings should I rule out first?
Check the pixel format before touching optics: a 12-bit camera whose data is captured or displayed as 8 bits without rescaling shows a washed-out, low-contrast image even on a bright target, and the fix costs nothing.
Put a uniform bright target under your lights and read the raw histogram. A white surface that should sit near the top of the range but peaks near 63 of 255, or gray values that step in fixed repeating bands, means a 10- or 12-bit stream is landing in an 8-bit path with the wrong scaling. Set the acquisition pixel format to the camera's native output, or rescale it there, and the contrast returns at once. The mechanics are in ADC bit depth vs pixel format.
Then seat the histogram before judging contrast. Check the black-level pedestal, which often ships at a different default on a new camera, and give the smaller pixels their light back: a 1.6 µm site collects (2.2/1.6)² = 1.89× less charge than a 2.2 µm one, so add about 0.9 stop of exposure or illumination.
How much contrast does diffraction leave at my pixel pitch?
Compute two frequencies and divide: the sensor's Nyquist frequency is 1/(2p) and the lens diffraction cutoff is 1/(λ·N), so the 1.6 µm pixel at f/4 samples 313 lp/mm against a 455 lp/mm cutoff and gets at most 20% contrast.
Both formulas come from geometry, not price:
Worked example — the swap described above, 2.2 µm down to the finer pitch on a 1/4-inch format at f/4, green light λ = 0.00055 mm:
- Old Nyquist: 1 / (2 × 0.0022 mm) = 227 lp/mm
- New Nyquist: 1 / (2 × 0.0016 mm) = 313 lp/mm
- Cutoff: 1 / (0.00055 mm × 4) = 455 lp/mm
- Old pixel: x = 227 / 455 = 0.50 → MTF = 39%
- New pixel: x = 313 / 455 = 0.69 → MTF = 20%
Same lens, same light, same scene: the modulation available at the sampled frequency halved. These pitches are typical for 2 MP and 5 MP parts on this format; put in your sensor's exact pitch and working f-number and the formulas give your own ceiling. The cutoff follows from the circular-aperture treatment of diffraction-limited systems, and the aperture-side consequences are worked through in diffraction limit and aperture effects. The Airy diameter, 2.44·λ·N, comes to 1.34 µm per unit of f-number at 550 nm — 5.4 µm at f/4, about 3.4 of the new pixels across.
Why does the same lens that was sharp on 2 MP run out on 5 MP?
A lens is built for a band of spatial frequencies, and a common design target asks for about 30% MTF at two-thirds of the sensor's Nyquist frequency — your old lens was chosen for 227 lp/mm, and 313 lp/mm was never in its brief.
What a given lens manages up there is written in its MTF curve; lens contrast and resolution shows how that curve is measured and read. Aberrations and diffraction pull in opposite directions as you stop down: wide open, aberrations soften the image; stopped down, diffraction takes over. On small pixels the diffraction side arrives early, which is why the aperture choice below matters more than it ever did at 2 MP — a lens comfortably sharp at f/8 on the 2.2 µm pitch is finished at f/8 on the finer one.
Does opening the aperture fix it, and what does it cost?
Yes — every stop you open multiplies the diffraction cutoff by 1.4, taking the finer-pitch example from 0% ceiling contrast at f/8 to 20% at f/4 and 41% at f/2.8 — but halving the f-number roughly halves the depth of field.
The ceiling values at 550 nm, perfect lens:
| f-number | Diffraction cutoff | Ceiling MTF at 227 lp/mm (2.2 µm pixel) | Ceiling MTF at 313 lp/mm (1.6 µm pixel) |
|---|---|---|---|
| f/2.8 | 649 lp/mm | 56% | 41% |
| f/4 | 455 lp/mm | 39% | 20% |
| f/5.6 | 325 lp/mm | 19% | about 1% |
| f/8 | 227 lp/mm | 0% | 0% |
Real designs land below these numbers, by more as the lens gets cheaper, and many FA lenses are cleanest between f/2.8 and f/5.6 rather than wide open — test the actual stop on a target before committing. Lock the iris once set; lens locking screws covers the hardware. The light-gathering case for opening up at fixed focal length is made for M12 lenses, and it carries over unchanged here.
Check the depth you would give away before turning the ring:
When is a new lens the only real fix?
When the cell cannot spend depth of field or light — calibration stations, gauging, cramped enclosures — the lens must carry the higher frequency band itself, which is why lenses are specified and filtered by the camera resolution they support.
The catalog sorts fixed-focal and telecentric lenses by supported resolution and by sensor format, so match the class to the camera instead of reusing whatever is on it now. The lens finder filters on both fields in one pass. Three fixed-focal catalog entries, one per resolution class, show the ladder:
| Specification | LS-FFL-0679 | LS-FFL-0602 | LS-FFL-0567 |
|---|---|---|---|
| Format | 2/3" | 1" | 1.1" |
| Resolution | 6.0 Megapixel (2.4 μm pixel size) | 10 Megapixel | 20 Megapixel |
| Focal length | 35 mm | 8.5 mm | 16 mm |
| Aperture | F2.8 Manual iris | F2.4 Manual iris | F2.8-22 |
| M.O.D. | 0.2 m | 0.1 m | 0.1 m |
| Size | dia. 33 x 30.6 mm | dia. 80 x 67.1 mm | – |
| Weight | 50 g | 270 g | – |
| Filter thread | – | M77 x P0.75 | M35.5 x P0.5 |
| Price | $219.00 | $429.00 | $679.00 |
All 3 share: Mount C-mount. Values as printed in each product's data sheet; open a part for its drawing and full specifications.
For the finest detail in that row, the 16 mm LS-FFL-0567 covers 1.1-inch sensors at F2.8-22 with under 0.3% distortion. Its wide-open stop is the kind of aperture budget a small pitch needs, and the higher MP rating exists precisely to hold contrast at those pitches' frequencies.
If the camera changes too, this is the class of sensor to shortlist — the MV-CS200-10UM pairs a 20 MP, 1-inch mono sensor with USB3 on a C-mount:
- Sensor format
- 1"
- Lens mount
- C-mount
- Interface
- USB3
- Resolution
- 16-25 MP
Can gain, black level or a LUT stretch the contrast back?
No: gain, black level and lookup tables act on numbers the sensor has already captured, so gain brightens noise exactly as much as signal and a LUT stretches flatness without adding any modulation the lens failed to deliver.
Each still has a job. Black level seats the noise floor at zero so the dark end does not sit milky; a LUT maps the camera's bit depth into the display range; gamma shapes the picture for an operator. None of them recreates modulation that never reached the sensor: a LUT stretch of a barely-modulated frequency yields a noisier gray image with a flatter histogram, because the contrast per pixel was fixed at exposure time by photons and optics.
Which fix should I choose for my case?
Decide by what the cell can spend: open the aperture when depth is spare, buy a resolution-matched lens when it is not, and use 2×2 binning when you mainly want the old camera's behavior back.
- Histogram banded, or white plateauing near 63 of 255: fix the bit mapping first. It costs nothing, and it is the only item on this list that is always right to try.
- Bright scene, depth to spare: open two stops, f/8 to f/4. In the worked example that lifts the new pixel from 0% to 20% and the old pitch from 0% to 39% at their Nyquist frequencies.
- Fixed depth, metrology duty: keep the f-number and order the lens in the camera's resolution class for the same format.
- Mostly want the old behavior back: 2×2 binning doubles the pitch to 3.2 µm, drops the Nyquist to 156 lp/mm where f/8 still leaves 20%, and gathers 4× the charge per binned pixel — at the cost of three-quarters of the pixel count.
Checks before you order the replacement lens: image circle at or above the sensor diagonal; mount matched to the camera; resolution class at or above the camera's; a minimum f-number low enough for your pitch's diffraction budget. Keep the focal length and the field of view holds too, because the format is unchanged — a format change would shift it, as same lens on a different sensor: how HFOV changes shows.
The camera finder and the configurator pair pitch, format and mount in one pass, and the lenses topic page collects the companion guides. If the histogram is seated and the ceiling table still leaves the detail flat, bring the pitch, format and f-number to the contact page and we will run the MTF numbers for your lens.
Questions
Does a smaller pixel always make the image worse?
No. Small pixels deliver more usable detail when the lens keeps useful contrast up to the new Nyquist frequency and the aperture stays reasonably open. A 1.6 µm pixel at f/2.8 behind a 20 MP-class lens resolves far more than a 2.2 µm pixel behind a lens chosen for 2 MP. The gray look appears when the pitch shrinks while the lens and f-number stay frozen; change one of the two and the extra resolution pays for itself.
Why does lost contrast read as gray rather than as blur?
Contrast and sharpness are one measurement, modulation. As modulation at a detail's frequency falls toward zero, its bright and dark lines converge toward the background gray. At 20% the pattern is present but faint; at 0% it is uniform gray. Because mid-band contrast fades gradually, the histogram flattens and the eye reports the image as grayer long before it reports blur.
Can I just add more light instead of touching optics?
Light fixes brightness and shot noise, not optical contrast. After the swap, lengthen the exposure or raise the illumination by about 0.9 stop so each small pixel regains the charge it lost, then rejudge on a seated histogram. If fine detail is still flat with black at the floor and white at the top, the loss sits at the lens and no amount of light recovers it.
Does the same limit apply to Bayer color sensors?
Yes, and it arrives earlier. A Bayer color camera splits the sites between green, red and blue, so each color channel is sampled more coarsely than the nominal pixel count suggests, and demosaicing then interpolates across the contrast that was lost. Treat a small-pixel color camera as needing roughly one resolution class better lens than the mono numbers indicate; the symptom is dull, desaturated fine detail.
How do I measure the MTF my own lens delivers?
Shoot a slanted-edge or line-grating resolution chart at your real working distance and aperture, in mono, with the histogram seated. Compute modulation from the edge profile, or note the finest grating still visible in lp/mm. Compare it with the Nyquist frequency of your pitch: if contrast fades near two-thirds of that value, the lens is the limit; if it fades well earlier, look to aberrations, dirt or a damaged element.