DO's and DON'ts
In the table below you find an overview and summary of the different aspects of DO's and DON'ts regarding figure creation and good scientific practice (GSP)



Editing Method
Usage Description
Suitable file types for raw images
- Always use the original file type from the software used for imaging
- Suitable alternative would be TIF but you might loose metadata and convert intensities to colors
- DO NOT use .jpg, .jpeg and similar file formats since they destroy all your data
Saving images of 8- to 32-bit type
- All intensity based images, such as fluorescence, Western Blots, EM, AFM, medical imaging
Saving images as RGB type
- Only image from photo-cameras, non-fluorescent immuno-histochemistry.
- DO NEVER convert or save intensity-based images as RGB type
Bit-depth conversion
- Avoid if possible and if necessary check that the image does not optically change.
- NEVER do before intensity measurements.
LUT changes
No problem (as long as the image is not converted into a color image)
Multi-color LUTs
- Choose wisely to not optically eliminate information and use a calibration bar to allow color interpretation
Channel Color Choices
- generally avoid blue and specifically combinations of red and green
- 2 channels: green + magenta
- 3 channels: cyan + magenta + yellow (careful with over-saturation)
- DO NOT merge more than 3 channels, rather make separate merges with 2 or max. 3 channels
Uneven Lighting - Flat-Field Correction
- Optimal with a real background image, taken during the imaging session.
- For fluorescent images use a autofluorescent plastic slide (e.g. from Chroma)
- Usable and necessary for intensity measurements.
Unspecific Signal Reduction
- The sliding paraboloid-based method described here (a simplistic deconvolution) can optically improve images in presence of diffuse unspecific signal. It partially reduces unspecific staining, uneven lighting and auto-fluorescence. Careful, it is based on an estimation and therefore not quantitative.
- DO NOT USE before intensity measurements!
- ALL images that need to be compared to each other need to be changed exactly the same
Contrast Adjustment
- Brightness adjustment is seldomly helpful
- For figures ONLY use linear-methods. AVOID gamma adjustment, histogram equalization
- Auto contrast CANNOT be used on multiple images which need to be compared to each other
- ALL images that need to be compared to each other need to be changed exactly the same (Use the Optimizer tools in Fiji)
- In color images, contrast should only be applied to the intensity information, not to color-tone (hue) or saturation.
- DO NEVER use before pixel intensity measurements
Image Scaling
- Check metadata for correct information about scaling
- Scaling is necessary before adding scale bars in Fiji, spatial measurements and 3D reconstructions
- DO NOT manually measure an existing scale bar to scale other images (this is only the last resort, if there is not information available)
Image Transformation
- DO NOT rotate images in pixel-based software (such as Fiji or Photoshop)
- DO NOT reduce image size in pixel-based software
- DO NOT use interpolation for transformations whenever possible
- DO transformations mainly in vector graphic-based software without interpolation
- Upscaling (making images bigger) MUST always be done by integer factors in pixel-based software
Image file types for figures
- Whenever possible use vector graphic supporting formats such as SVG, PDF, EPS
There also exist checklists that should help authors to orient on the specific parts of an experiment and image analysis that should be reported for transparency as well as reproducibility reasons: