A conservation volunteer photographs a large moth on a garden wall after sunset. The RAW frame is noisy, but the wings have a recognizable row of pale spots and the antennae are thin enough to disappear in a heavy cleanup. The practical goal is a legible identification photo, not a dramatically textured image that might imply a different species.
DxO PureRAW 6 documents DeepPRIME XD3 for RAW noise reduction and detail extraction on supported Bayer and X-Trans cameras. This article designs a hypothetical processing and review workflow. No moth image was processed for this guide, and the drawn before-and-after examples are teaching diagrams rather than claims of measured enhancement.
Start with the original RAW and an identification note
Keep the camera-native RAW file, any untouched camera JPEG, the capture time and a short location note in a read-only source folder. Before looking at enhanced detail, record the visible shape of the antennae, the number and placement of obvious wing spots, and any soft areas that cannot be resolved.
If the moth was not identified by a qualified observer, leave the species field provisional. Cropping and noise reduction are presentation choices; they are not independent biological evidence. Retain the full frame so an editor can revisit the moth’s orientation and surroundings.
Check camera, file and processing compatibility first
DxO says PureRAW 6 supports RAW files from compatible cameras, including supported X-Trans models, but not every DNG workflow. The official settings guide specifically excludes certain merged, compressed and smartphone DNG files.
Confirm the camera and lens combination in DxO’s supported-module lookup and check the operating system, RAM and GPU requirements before scheduling batch work. Processing a converted JPEG is not the same as reconstructing RAW sensor data. If a source file is unsupported, do not silently substitute an unrelated file format just to demonstrate the feature.
Compare DeepPRIME 3 and XD3 on one difficult crop
Open a duplicate job containing the same high-ISO source frame and select a small region with both a pale wing spot and a smooth, shadowed wall. DxO lists DeepPRIME 3 as a fast, high-quality choice and DeepPRIME XD3 as the more aggressive detail-extraction option.
Generate separate previews rather than changing modes repeatedly with no record. Name the outputs by processing choice, not by an invented quality score. The best result is the one that preserves defensible visual information, including subtle color transitions around the spot, not automatically the sharpest-looking image.
Use luminance and Force Details cautiously
DxO describes a Luminance setting that trades visible grain against smoothing, and Force details, which changes the threshold for extracting texture. Force details defaults to zero; pushing it right can produce a less natural rendering. Begin near the default and compare a modest adjustment with the baseline at matched zoom.
Inspect the wing edge and antenna tip, which may break or develop crisp halos under overprocessing. Avoid presenting a freshly created highlight as a scientifically confirmed marking. Record each adjustment so a reviewer can tell processing artifacts from source structure.
Distinguish useful texture from a made-up pattern
A risky hypothetical crop adds a convincing extra white dot next to the outer wing edge. An accurate editorial decision would reject that candidate unless the dot is also visible in the original capture or a separate, legitimate reference. View the raw and processed images side by side at 100 percent and at a normal viewing size.
A spot that appears only in one enhancement is not dependable. Fine hair and wall grain can also be sharpened into false stripes. The comparison diagram illustrates the visual logic; it does not claim that XD3 generated a real error.
Export a reviewable file without hiding its provenance
Use a supported output format appropriate to the next editor, keep the original RAW unchanged and attach a processing note alongside any exported DNG or TIFF. Avoid flattening a high-contrast color grade into the master identification file.
Give the photographer a version with a meaningful suffix such as “noise-reduced-for-review” while retaining a source filename map. At export, check the full-frame dimensions, image rotation and color interpretation. If a stronger version may be used as an illustrative print, label it as enhanced, and do not use the result to establish new morphology.
Build a four-point visual audit around the wings
Review the left and right wing patterns independently, compare the antenna and leg edges, check whether wall texture has turned into repeated dots, and verify that color patches remain consistent with the original. A simple overlay can reveal invented edges or lost small features, but it is not a substitute for source inspection.
Ask a second person to inspect the source crop before showing them the enhanced candidate, reducing the risk of accepting the processed shape as a memory of the original. Archive both opinions when they differ.
Publish as an enhanced photograph, not a new observation
Before releasing a nature-center handout, decide whether the processed picture is suitable for identification, illustration only, or neither. The caption should say when the image has undergone computational noise reduction and must not claim a newly resolved pattern or species determination that the source cannot justify.
Check any accessibility description against features genuinely visible in the original. Never replace the source in a research repository. DxO can assist with a difficult high-ISO file; the documentation does not guarantee reliable recovery of every tiny biological detail.
How We Research
We prepared this guide on November 4, 2026 by reviewing developer documentation: DxO PureRAW 6 release highlights; DxO processing and denoising controls; DxO PureRAW input and hardware settings. We checked what each named function can do, its documented plan or format constraints, the available controls and the documented review and handoff points. The project examples are hypothetical teaching scenarios, not real software trials, customer records, measured benchmarks or evidence of a particular output. The original diagrams show the risks editors should inspect; they do not reproduce or impersonate product screenshots. For a live assignment, verify the current feature availability, rights, security requirements and output against the real source materials, then require a named human to authorize release.
Frequently Asked Questions
DxO states that PureRAW 6 supports DeepPRIME XD3 processing for supported Bayer and X-Trans sensors. Camera and RAW-format compatibility still need checking.
Not necessarily. Compare previews against the untouched RAW, especially where small markings and antennae could look artificially sharp or become smooth.
DxO describes it as a detail-extraction threshold control, defaulting to zero. Higher settings may produce less natural-looking texture.
No. An enhanced image is not independent observation; identification must rely on verified source evidence and appropriate expertise.
Retain the original RAW, its metadata, the processing settings and a clearly labeled processed file so another editor can inspect any claimed detail.