vallejo
Offline
• • •
Upload & Sell: Off
|
Here is what Grok says:and now we’re at PL9 and DeepPrime XD is not on beta since late version 8.
DeepPRIME XD fully optimized in PL8, but beta XD3 (as of mid-2025) initially excluded 40MP files—full support arrived in a June 2025 update
### DXO PhotoLab Optimization for Fuji X-Trans Files: 26MP vs. 40MP Sensors
DXO PhotoLab has supported Fujifilm's X-Trans sensors since version 5 (released in 2021), initially for basic demosaicing and noise reduction, with advanced AI-based tools like DeepPRIME and DeepPRIME XD added in subsequent updates (e.g., version 6.4 in 2023). This applies to both APS-C X-Trans sensors (like those in the X-T4/X-T5 at 26MP and X-H2/X-T5 II at 40MP) and medium-format GFX X-Trans sensors (e.g., GFX100 at 102MP, which uses a similar layout but at higher resolution). Optimization in PhotoLab primarily involves demosaicing (interpreting the non-standard X-Trans color filter array), noise reduction (via DeepPRIME/XD), lens corrections, and sharpening to minimize common X-Trans artifacts like worming (false color patterns) and moiré.
As of November 2025 (with PhotoLab 8 and the beta DeepPRIME XD3), there **are some differences** in optimization between 26MP and 40MP X-Trans sensors, driven by pixel density, file complexity, and software maturity. These aren't drastic for most users but can impact workflows for high-ISO, cropping-heavy, or large-print scenarios. Below, I'll break it down based on sensor characteristics, PhotoLab's handling, and user-reported outcomes.
#### Key Sensor Differences Affecting Optimization
- **Pixel Density and Detail**: The 40MP sensor (X-Trans V, ~1.0μm pixel pitch) packs ~55% more pixels than the 26MP (X-Trans IV, ~1.3μm pitch) into the same APS-C area. This increases resolving power (up to ~24% theoretical gain in linear resolution) but amplifies noise visibility at 100% zoom and demands finer demosaicing to preserve micro-details without artifacts.
- **Noise Profile**: Smaller pixels on 40MP files can appear noisier when viewed pixel-peeped or cropped heavily, though they match or outperform 26MP at matched output sizes (e.g., screen or print). X-Trans's layout exacerbates this, requiring stronger AI denoising.
- **File Size and Processing Load**: 40MP RAWs are ~50-60% larger (~80-100MB vs. ~50-60MB for 26MP), leading to longer render times in PhotoLab (2-3x on average, depending on hardware).
#### PhotoLab-Specific Optimization Differences
PhotoLab treats both resolutions similarly in its core pipeline (RAW import, optical corrections, and export), but advanced features highlight variances:
| Aspect | 26MP (e.g., X-T4) | 40MP (e.g., X-H2) | Key Implications |
|--------|-------------------|-------------------|------------------|
| **Demosaicing & Basic Support** | Full support since PL5; handles X-Trans IV layout reliably with minimal worming. | Full support since PL5 for X-Trans V; similar reliability, but higher density can reveal subtle moiré in fine patterns (e.g., fabrics) if not using XD mode. | No major gap—both use the same X-Trans engine. Users report "excellent" detail extraction on both. |
| **Noise Reduction (DeepPRIME/XD)** | DeepPRIME XD fully optimized; excels at ISO 6400+ with natural detail retention (e.g., fur/feathers). | DeepPRIME XD fully optimized in PL8, but beta XD3 (as of mid-2025) initially excluded 40MP files—full support arrived in a June 2025 update. May require more aggressive settings for equivalent cleanliness. | 40MP benefits more from XD3's improved demosaicing, reducing artifacts in crops. At matched sizes, 40MP often looks "materially cleaner" post-processing. |
| **Sharpening & Detail Enhancement** | Standard settings suffice; "Detail" slider at 20-40 for balanced output. | Requires higher "Detail" values (40-60+) due to density; unsharp masking is more critical to counter softness in downscaled exports. | 40MP demands tweaks for optimal sharpness—e.g., lens-specific modules (now covering most Fuji XF lenses) perform better here. |
| **Processing Time & Stability** | Faster (e.g., 10-20s per file on mid-range CPU); fewer bugs reported. | Slower (20-60s per file); early XD3 betas had grayed-out options for 40MP, now resolved. | 40MP workflows may need batching or GPU acceleration for efficiency. |
| **Output Quality (e.g., Crops/Prints)** | Great for moderate crops; noise comparable to full-frame at base ISO. | Superior for heavy cropping (e.g., wildlife); but raw files may show +10-20% more luminance noise at high ISO without XD. | 40MP shines in PhotoLab for landscapes/large prints, with users noting "extra pop of sharpness." |
#### User Experiences and Recommendations
- **Strengths for Both**: PhotoLab is widely praised as one of the best for X-Trans (rivaling or beating Lightroom/Capture One in noise handling), especially with DeepPRIME XD for high-ISO recovery. Fuji users report it "transforms" underexposed shots (e.g., pushing ISO 6400 by 2 stops yields cleaner results than Adobe).
- **40MP-Specific Notes**: Early adopters (e.g., X-H2 users) noted slight delays in full AI support, but by 2025, it's seamless. For optimization, start with "DxO Optical Corrections only" preset, then apply XD at default strength—adjust luminance to 40-50 for 40MP to match 26MP noise levels. Avoid over-sharpening to prevent halos on dense files.
- **No Differences?**: If you're not cropping heavily or printing large, the gap is negligible—many users see no IQ edge from 40MP after downscaling.
- **Tips**: Download the 30-day trial to test your files. For GFX 40MP (X-Trans CMOS II, similar layout), support is identical to APS-C 40MP, with even better lens modules.
In summary, while the core optimization is consistent, 40MP sensors require more parameter tuning (e.g., sharpening, denoising strength) and briefly lagged in beta features—but PhotoLab handles them exceptionally well now, often highlighting the resolution advantages over 26MP. If you have specific files or cameras in mind, I can suggest targeted settings.
Rand47 wrote:
I don’t think DXO Pure Raw supports the Fuji X-Trans 40 megapixel raw files (yet). It will use one of its options to demosiaic them, and it’s better than LrC default, but not optimized for X-trans, I don’t think.
Most people don’t understand (DXO has done a poor job of marketing) that Pure Raw is not “just a noise handling plug-in.” It goes in and copies the original underlying raw file, and takes it to DXO Pure Raw where it uses its own demosaicing routine, lens correction modules, sharpening, noice handling, and most impressive, its detail extraction slider that has the unfortunate name of “force details.” The results are excellent and where noise handling is concerned (a small part of what DXO does) it is superior in all my tests agains both Adobes new AI Noise, and Topaz.
The only down side to DXO Pure Raw is if you have unusual lenses that they have not made lens correction profiles for. E.G. my two GF Tilt-Shift lenses. It also took a long time for them to do Modules for lenses like the Nikkor Nocht 58mm f/0.95. But over time their inventory of profiles is getting broader.
The other down side is that it is slow, so not viable to run all your captures through it as a routine workflow. But, it has an amazing feature for your “worked up” keepers in LrC. It takes note of your edits in LrC and reapplies them to the new DNG that it produces when it returns them to LrC. The difference in rendered detail is significant, and all my files that I am going to print are run through DXO Pure Raw and then fine tuned as necessary.
It is an amazing program, little understood, IMO. I have both GFX and X-Trans 40MP cameras and I can’t wait for DXO to release an update for the 40 MP sensor.
Rand...Show more →
|
