Comments (1)
Hi, @0Error0Warning, thank you for your interest about my work.
I saw that issue in my repo, but I’ve been quite busy these days to have time to answer the issue.
I am sorry about that :(
I think that there is a slight difference between two visualization process.
In 1_make_mixture_map.py, the white balance process is done by following steps.
- Open raw image using rawpy (Directly open RAW file, not using DCRAW or Tiff)
- Apply WB using illumination map, directly accessing to raw_image_array in RawPy instance (apply_wb_raw function, line 194)
- Visualize sRGB image using postprocess function
So [open RAW → direct WB application → postprocess]
In visualize.py,
- GT tiff file is generated through [ Tiff conversion using DCRAW → WB application → saved as Tiff ]
- To visualize tiff file into sRGB, we utilize fake raw file and copy pixels into RAW instance
- Visualize sRGB image using postprocess function
So [Tiff conversion using DCRAW → WB application in Tiff → overwrite pixels into fake RAW instance → postprocess]
You can compare the differences in these processes and explore where the differences in values occur.
You can compare the value of raw_image after opening the RAW image with Rawpy with the value saved as a TIFF file using DCRAW.
However, sRGB images are images that have undergone a lot of post-processing, so if you use images that are visualized according to one criterion, there should be no problem.
For white balance research, the sRGB image is for reference only. If you want to compare the correct values, please use the values in the RAW domain.
The pixel values in the dataset I provide are all provided in TIFF format using DCRAW, so they will have consistent values.
from lsmi-dataset.
Related Issues (20)
- MAE_rgb and MAE_illum HOT 1
- Sony dataset HOT 1
- The blacklevel confusion of Sony sensor & Missed saturation clipping after wb HOT 7
- tint_map has negative values resulting in NaN in uvl conversion HOT 1
- illum_map becomes ZERO in 2_preprocess_data.py due to alpha and chormacities HOT 1
- train val test split for single illuminant and mixed illuminant training schedules HOT 3
- #
- HDRNet HOT 3
- No access to dataset HOT 1
- The problem of Sony-sensor's effective bit-depth
- about pre-trained model HOT 2
- about pick data HOT 1
- about train size
- about data collection HOT 1
- bug in preprocess_data.py HOT 3
- Error in Visualization part HOT 1
- json files for nikon and sony subsets HOT 1
- SAMSUNG S23 ultra data HOT 6
- Error in the meta,json of sony HOT 5
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lsmi-dataset.