Save a thermal colormap TIFF for reliable viewing

A thermal image can look correct inside an acquisition application yet fail to open after saving. The display may be showing colors generated from a grayscale array, while the TIFF writer receives something different: measurement values, palette indices, or RGB pixels. This distinction matters when you need both dependable viewing and reusable inspection data. Choose the saved representation first, then verify the file independently of the application that created it.

Choose measurement values or rendered colors first

For reliable viewing, save an explicitly rendered RGB TIFF. For measurement or later analysis, retain the original scalar image separately, together with the information needed to interpret its values. One file does not automatically satisfy both purposes.

A colormap is a lookup table that assigns colors to values. Changing the display colormap does not necessarily change the underlying image array or the format passed to the writer.

These are three different representations, even when they appear identical on screen:

The detailed workflow for preserving high-bitdepth thermal grayscale images addresses the measurement branch. Here, the decision is how to save the displayed colors without confusing them with measurement values.

Understand the mapping between values and colors

A typical display pipeline scales a scalar value into a palette index, then retrieves a color from that palette. The following example uses a palette containing 256 colors and zero-based indices:

Here, x is a valid scalar pixel value, a and b are the display limits, and b must exceed a. The function clamp limits the normalized result to the interval from 0 to 1. This is one explicit mapping convention; an application may use different rounding or interpolation.

Suppose an already calibrated image uses display limits of 20 °C and 80 °C. For a pixel measuring 35 °C:

That pixel receives the color at zero-based palette index 64. The color itself depends on the palette; there is no universal RGB value for that temperature.

Values outside the display limits receive endpoint colors in this example. Invalid measurements should follow a separate masking rule rather than silently becoming valid-looking temperatures.

Automatic contrast adjustment can change the display limits between frames. The same measurement may then appear in different colors. The guide to brightening grayscale images without losing highlight detail explains why display adjustments should remain separate from preserved data.

Build an explicit RGB export pipeline

An RGB export should turn the intended appearance into actual color pixels before writing the file. Do not assume that a display-window setting accompanies a saved grayscale array.

Use this sequence:

1. Keep the acquired scalar array unchanged. Make a separate working copy for display processing so later adjustments cannot overwrite the source.
2. Choose the display limits. Use fixed limits when colors must remain comparable across frames. Record any automatic limits used for individual images.
3. Handle invalid values. Define how missing, saturated, or otherwise unusable samples appear. Preserve the validity mask if analysis requires it.
4. Apply the scaling and palette. Generate an array containing explicit red, green, and blue channel values for every displayed pixel.
5. Convert the rendered array to the intended sample type. An unsigned 8-bit RGB image is a practical starting point for broad viewing compatibility.
6. Write the array as RGB TIFF. Do not pass an additional palette argument when the array already contains RGB channel values.
7. Read the saved file back. Check its dimensions, channel count, numeric type, and appearance before adding the export to an automated workflow.

A screenshot can capture the visible appearance, but it may also capture interface elements, resized pixels, or overlays. Direct RGB rendering gives you control over image dimensions and which annotations are included.

If you need a legend, create a separate annotated presentation export or deliberately add the legend to the rendered image. Record the display limits and units alongside it. A color bar that no longer matches the exported mapping can mislead a reviewer even when the TIFF opens correctly.

Use indexed TIFF only when the receiving workflow supports it

Indexed TIFF is not inherently defective. It is a different representation with additional relationships that must remain consistent.

For a conventional 8-bit palette image, each pixel selects one of 256 palette entries. A writer must encode the image as palette color and include a compatible color table. A reader must interpret both correctly.

Do not simply attach a palette to an arbitrary grayscale array and assume its values become valid indices. Depending on the writing API, integer and floating-point inputs can follow different index conventions. Some APIs use zero-based indices; others expose one-based indexing or normalized values. Follow the writer's documented input contract.

Converting grayscale values to palette indices is also different from converting an existing RGB image into indexed color. The latter may choose a new palette or introduce dithering. That can be useful for presentation graphics, but it is unnecessary when you already have scalar values and a defined display palette.

Choose indexed TIFF when the downstream software explicitly supports it and your round-trip test passes. Choose RGB when avoiding palette interpretation is more valuable than reducing the pixel-array size.

Verify the file structure, not just its thumbnail

A missing file-browser thumbnail is not proof that the TIFF is corrupt. Preview handlers often support fewer TIFF variants than dedicated image readers. Conversely, a familiar-looking thumbnail does not prove that the full-resolution data is correct.

Inspect the saved file with a reader that reports TIFF structure. For a straightforward RGB viewing export, check these expectations:

Compare the read-back RGB array with the RGB array supplied to the writer. For a lossless path without color conversion, their sample values should match exactly.

Check a dark pixel, a bright pixel, a midrange pixel, and an invalid-data location. This catches channel-order mistakes, clipping, and incorrect masking more efficiently than inspecting only the overall appearance.

If software is examining the file as a byte buffer, distinguish the encoded container from decoded pixels. The explanation of image bytes versus pixel data helps identify errors at that boundary.

Estimate storage without confusing channels and precision

RGB does not always require three times the storage of your original image. The ratio depends on both channel count and sample depth.

Consider an illustrative image measuring 320 × 240 pixels. Before compression, headers, or metadata:

An 8-bit indexed pixel array would contain 76,800 bytes, plus its palette and other TIFF structures. Actual file sizes depend on compression, image content, tags, and additional pages.

This comparison explains a practical trade-off: RGB spends storage to make each pixel's color explicit. It does not preserve the original scalar precision simply because it contains more total bytes.

Keep the export decision separate from camera selection

The same representation problem can occur with different camera interfaces and spectral bands. A camera's connection type does not determine whether your application writes grayscale, palette, or RGB TIFF.

The LS-IRC-0044 is a thermal camera with a GigE interface. Its catalog identity does not, by itself, establish which TIFF representations an application can export.

The LS-IRC-0016 is an LWIR thermal camera module with a USB3 interface. Confirm the acquisition software's available arrays and export controls separately from the hardware interface.

When comparing infrared cameras, ask what the acquisition path actually delivers: display pixels, sensor counts, calibrated values, or separate streams. Use the camera finder for hardware filtering, but request documentation for the required data output before choosing a system.

The distinction also applies to hyperspectral cameras. A selected band can be displayed with a palette, while a deliberately constructed composite can use RGB channels. Neither rendering preserves an entire spectral dataset.

The LS-HSC-0004 is an area scan hyperspectral camera. It is a different acquisition category, not a substitute selected merely because its images can also be colorized.

For the underlying distinction between bands and color presentation, read how hyperspectral imaging works.

Diagnose failures in a controlled order

Change one export choice at a time. Start with a single-page, uncompressed, unsigned 8-bit RGB test image to reduce the number of variables.

If the simple RGB test opens but the original export does not, add compression or other features back individually. If neither opens, test a small known image through the same writer before investigating camera acquisition.

Do not treat reverse color lookup as measurement recovery. Multiple source values can share the same rendered color, and clipping removes distinctions entirely. The explanation of thermal image color-to-temperature mapping describes that limitation.

Check the handoff before standardizing the workflow

Before deploying the export routine, confirm the following:

• Purpose: the recipient knows whether the file is for viewing or quantitative analysis.
• Representation: the saved array and TIFF interpretation agree.
• Mapping: palette, display limits, and invalid-data treatment are recorded.
• Compatibility: an actual receiving application opens the completed file.
• Round trip: decoded pixels match the intended export array.
• Preservation: original measurements remain separate from contrast adjustments and annotations.
• Association: the presentation image and source data share an unambiguous capture identifier.

Keep calibration and acquisition information with the measurement record. Use the guidance on thermal image metadata export when deciding what supporting information must travel with an image.

For a new acquisition system, use the camera configurator to define the hardware requirements. If the export contract remains unclear, contact the applications team with the required scalar output, viewing format, and receiving software requirements.