Images with high image quality are instantly pleasing to the eye.[1]


If you are interested in high-quality rendering, you might wonder if the images displayed on your CRT, LCD, film, or paper will result in light patterns that are similar enough to the "real world" situation so that your eye will perceive them as realistic. You may be surprised to learn that there are several steps in the digital image creation pipeline where things can go awry. [2]




What does image quality mean when talking about 2D static images (computer graphic, non-photographic digital images), especially fractal images?

Factors that affect quality include all elements of the chain:[3]

Tips

edit

How to make better images

edit

How to make better images[5]

How to create realistic 2D static images (Mostly by Andy Ritger)[6]

  • Higher than standard (SD) pixel resolution (HD, UHD, "4K" and "8K")
  • Higher than standard bit depth per colour channel (Now standard is 8 bit depth): So 10 bit is a new target (See also number type and precision)
  • Wider Colour gamut (WCG): Express a wider range of colours than today (BT.2020 colour gamut)
  • High Dynamic Range (HDR): express a wider range of luminance than today (see AVIF file type)

Tips (By Paul Bourke)

  • Do fractal creation in 16 bit, not 8 bit, integer or floating numbers for each pixel calculation
  • Apply anti-aliasing by Supersampling each final pixel (Render a 30,000 pixel version of the image)
  • Do all Colour external to fractal creation using gradient maps, allows you to make appearance decisions independent to the creation

Tips (By 3DickUlus)

  • "Using 100 samples per pixel, so rendered the full image 100 times with small random jitter (Movement at sub pixel size) and accumulate the result, gives really nice anti-aliasing". 3DickUlus

Tips (By quaz0r)

  • Use normalised floating-point values: Values between 0 and 1
  • Use an image I/O library that can take these values as input and output to any format
  • Average samples together internally in your program, instead of using an external program
  • Poisson-disk sampling (Blue noise) is the best sampling pattern

Check all elements in the chain from the application to the monitor if they are ready:

What makes image poor quality?

edit

What makes image good quality?

edit
  • Looks good and is instantly pleasing to the eye
  • No signs of poor quality
  • Looks realistic
  • Should highlight features of the data, and not highlight features that are only in the gradient itself = Avoid distorting the data (If it's a scientific visualisation not fractal art)

How to check quality of the image?

edit
  • Visually, if it looks good and is instantly pleasing to the eye
  • Use a zoom function to check for distortion, blemishes

How to set up best viewing conditions ?

edit

web-browser-color-management-test

edit

How to make quality print of a fractal image

edit

Printing tips


How To Get My Monitor To Match My Printer ?[7]

Resolution and size

edit

Visual rating of the image resolution quality[8]

  • Good: Pixels are not visible without zoom
  • Poor: Pixels are visible without zoom (Pixelated)

Resolution: Image resolution (Measured in pixels per inch, PPI [9]) describes the detail an image holds. The term applies to raster digital images, film images and other types of images. Higher resolution means more image detail. (By Becky J. Ritchie)

See also:

Measures of resolution:[10]

  • DPI: Dots per inch. Used in printing and scanning and refers to the dots of ink on one line across one inch. It is a measure of spatial printing or scanner dot density. In particular the number of individual dots that can be placed in a line within the span of 1 inch (2.54 cm). Similarly, dots per centimetre (D/cm or DPcm) refers to the number of individual dots that can be placed within a line of 1 centimetre (0.394 in).
  • PPI: Pixels per inch. Describes display resolution. Used for display (web, monitors and video production). A pixel is a dot on a video screen
  • LPI: Lines per inch. Used for half-tones and is the measurement of how close together the lines are in the grid

How to compute PPI

Difference between size and resolution[11]

  • Image size (in pixels = horizontal resolution x vertical resolution) shows how much information is in the image
  • Image resolution is related to printing or display

Examples:

  • The final printed or displayed size needed is 300 dpi @ 5”x7”. Means the image should be 1500x2100 or larger
  • The final printed or displayed size needed is 300 dpi @ 8”x10”. Means the image should be 2400x3000 or larger

The quality of a printed image depends on two things:

  • The capabilities of the printer (DPI)
  • The original image resolution (Size)

The quality of a displayed image depends on:

  • The capabilities of the monitor (PPI)
  • Zoom factor
  • The original image resolution (Size)

Resolution types

  • Spatial (Size of smallest object that is recognisable in mm), maybe also divided into horizontal and vertical
  • Contrast (Minimal contrast between what is recognisable)
  • Temporal (Minimal time in ms, for moving, not static, images)

Compression

edit

Data compression

Bit depth

edit

RGB cube: 3, 6, 9, 15 and 18 bit:

     

Greyscale: 2, 3, 4, 6 and 8 bit:

     

What bit depth per colour channel (Precision and number type) should I use?[12]

  • So how many bits do you really need?[13]
    • How many bits do you need for sharing on the Internet? 8-bits
    • How many bits do you need for scanning? When a look at the transparency shows that I'm going to have problems with the shadows then I import 48 bit RAW images, since I want to maximize the advantage
    • How many bits do you need for printing? 8-bits is fine for the final output, but use 16 if your printer supports it
    • How does colour space impact bit depth? Colour space (Gamut) is related to bit depth.
    • How many bits should you use for image processing? Use 32-bits for processing an HDR file
      • What precision should you choose in GIMP for editing? Choosing 32-bit floating point precision allows you to take full advantage of GEGL's 32-bit floating point processing. 64-bit precision is made available mostly to accommodate importing and exporting very high bit precision images for scientific editing[14]
    • How many bits can you can see? A 10-bit gradient is OK for display (No visually detected banding)
    • Why use more bits than what you can see? To avoid numerical errors producing visual errors in the image during image editing (Image processing)

Tips to get 10 bit colour depth working on all elements in the chain from the application to the monitor needed to support it (By Ted Lansing). This includes:[15]

  • Hardware
    • Video card
    • Video connection
    • Monitor
  • Software
    • Application
    • Operating system
    • Video card driver

More description

Target destination

edit

Technical destination of the image:

  • For display (RGB)
    • Quality display (Full size image without compression or with lossless compression)
    • Preview (Small image with reduced size, maximum compression even if the quality is reduced)
  • For printing (CMYK)
  • For archiving (Full size image without compression or with lossless compression, with all colour space versions): EXR, TIFF

Application of the image

  • Artistic (Free art): Fancy visualisation, image should be interesting and pleasing to the eye
  • Design (Applied art)
  • Scientific (Scientific visualisation)
    • Colour gradient should highlight features of the data and should not highlight features that are only in the gradient itself
    • Colour gradient should be perceptually uniform. Uniformity: Perceived differences in colour must accurately reflect numerical data differences
    • Metric: Enable users to accurately read absolute values in data = Value reading tasks
    • Form: Gradient should enable users to correctly perceive the shape = Pattern perception
    • Order: ColourMaps must preserve the order in data
    • Separation: Different data must be perceived differently
    • Colour-vision-deficiency (CVD) friendly = Accessible to those with colour blindness
    • Printer friendly
    • Photocopy safe

Black

edit
  • Black (0,0,0,100) is not black; it is a dark gray. The reason it's gray, rather than black, is that the ink is partially absorbed by the paper …" and therefore the grayness of the key-black is relative to the properties of the paper: newspaper sucks a lot while glossy coated paper might not.
  • "built black" or "rich black." This is usually something like 60/60/40/100, or 40/30/30/100 [16]

White

edit

Graphic workstation

edit

Software

  • Imaging system
  • Colour Management System

HDR processing ( HDRI)

  • rendering graphic (HDRR)
    • HDR tone mapping
    • HDR encoding
  • display graphic



What do I need for HDR on PC?[17] Monitor with

  • A high maximum brightness, ideally 1000cd/m2, but at least 600cd/m2+ if not
  • At least 90% coverage of the DCI-P3 colour gamut
  • A 10-bit panel



Hardware

edit

Recommendation (By Matt Smith) for what you need to achieve great HDR on a PC

  • A monitor that's DisplayHDR 1000 certified or a quality HDR television.
  • A video card from the AMD RX 400, Nvidia GTX 900 series, or Intel integrated graphics found in Intel 7th-gen Core, or newer
  • An HDMI 2.0a, DisplayPort 1.4, or Thunderbolt 4 connection, or newer
  • HEVC High Efficiency Video Coding
  • HDR content, such as an HDR-compatible game or streaming service or HDR image (Photo or digital graphic)

Graphic workstation:

Scanner:

  • Canon: Colour 48-bit in/24-bit out, greyscale 16-bit in/8-bit out
  • "I scanned this transparency on my Imacon Photo scanner in 48 bit mode because I wanted to gather as much data to work with as possible. Once I finished adjusting Curves and Colour Balance I then converted to 24 bit for printing and web display". (By Michael Reichmann)[18]

Monitor 3 levels of monitor brightness:

  • A low brightness if they have a peak brightness below 300 nits
  • A standard brightness if they have a peak brightness between 300 and 350 nits (SDR)
  • A high brightness if they have a peak brightness above 350 nits (HDR)
  • Note that humans can detect brightness levels as low as 0.000001 nits and as high as 100,000,000 nits
How to calibrate your monitor
What is the reason for calibration?
  • If the screen and the printer are calibrated, the green in the leaf, and the photo, and the print will look the same.

Calibration targets:

  • white point
  • neutrality of greys
  • predictability and consistency of tone response


How to test your display (Monitor, projector) for HDR support

Monitor requirements

Ports and cables


Examples:

ASUS ProArt

  • Mouse: ProArt Mouse MD300 : Up to 4200 dpi sensor resolution and Up to 1000 Hz polling rate
  • Monitor: ASUS ProArt display PA32UCX-PK 4K HDR IPS Mini LED: World’s 1st HDR 1600 nits & 120 Hz monitor (3000 USD)
    • 32ʺ 4K 120 Hz monitor with true 10-bit colour
    • 1600 nits peak and 1000 nits sustained brightness
    • Delta E < 1 Colour accuracy / Hardware calibration
    • Dolby Vision®, HDR10 and HLG support
    • X-rite i1 Display Pro included
  • Motherboard: ProArt B660-Creator D4 (200 USD)


Apple 32-inch Pro Display XDR with Retina 6K Display - Standard Glass

  • 32-inch LCD display with Retina 6K resolution (6016 by 3384 pixels)
  • Pro Stand and VESA Mount Adapter sold separately
  • Extreme Dynamic Range (XDR)
  • Brightness: 1000 nits sustained, 1600 nits peak
  • Contrast ratio: 1,000,000:1
  • P3 wide color gamut, 10-bit color depth

Software

edit

Colour Management System (CMS):[22][23]

  • A device-independent colour space
  • ICC profiles for each device (i.e. printer, scanner, monitor, digital camera, etc.): Input and output profiles
  • A Colour Matching Module (CMM)
  • Colour workflow

Examples:

  • Linux_color_management
    • Little CMS - OPEN SOURCE small-footprint CMS, with special focus on accuracy and speed.
  • Apple’s color management system is called ColorSync
  • Microsoft’s is called Image Color Management

OS Settings

Web browser

  • chrome flags: Force color profile
    • brave://flags/

Libraries

  • Image I/O library that can take these values as input and output to any format. This can include different bit depths, HDR and different colour spaces
    • OpenImageIO = Reading, writing and processing images in a wide variety of file formats, using a format-agnostic API
  • Image processing
    • ImageJ: Data types: 8-bit greyscale or indexed colour, 16-bit unsigned integer, 32-bit floating-point and RGB colour

Image processing programs:

RAW file editors[24][25]

  • RawTherapee
  • Rawstudio
  • DigiKam
  • Darktable
  • Filmulator
  • Photivo
  • LightZone
  • Shotwell
  • UFRaw

Number precision and type in GIMP:[26]

  • Integer precision options
    • 8-bit integer
    • 16-bit integer
    • 32-bit integer
  • Floating point precision options
    • 16-bit floating point
    • 32-bit floating point

Image file type - High dynamic range file formats - 32‑bpc HDR images

  • AVIF is a file format developed by Alliance for Open Media. It's an open standard which already enjoys fairly broad support in major web browsers for standard images. AVIF offers numerous advantages over JPG
  • PNG has 1, 2, 4, 8 and 16-bit greyscale support and 8 and 16-bit-per-sample (That is, 24 and 48-bit) Truecolour support
  • PFM file - floating point image format for HDR images
  • EXR (OpenEXR): 16-bit OpenEXR files can be considered a good colour precision/file size compromise for the general case[27] OpenEXR has support for color depth using: 16-bit floating-point (half), 32-bit floating-point and 32-bit unsigned integer
  • TIFF
  • RGBE image format or Radiance HDR is an image format invented by Gregory Ward Larson for the Radiance rendering system. It stores pixels as one byte each for RGB (Red, green and blue) values with a one byte shared exponent. Thus it stores four bytes per pixel

Final note

edit

Now most of the hardware/software is using 8-bit color channels, some monitors / TV have 10 -bit


Abbreviations

edit

IPT components:

  • The I ("intensity") component is a luma component that represents the brightness of the video
  • CT : blue-yellow (named from protanopia) chroma components
  • P : red-green (named from protanopia) chroma components



Formal International Standards Bodies

edit
  • CIE - International Commision on Illumination
    • Division 1 : Vision and Colour
    • Division 2 : Measurement of Light and Radiation
    • Division 8 : Imaging
  • IEC - International Electrotechnical Commission
    • TC100/PT61966 - Color Management
  • ISO - International Standards Organization
    • TC6 - Paper, board and pulps
    • TC36 - Cinematography
    • TC42 - Photography
      • PMAI - US Technical Advisory Group
      • PIMA
    • TC130 - Graphic Arts
      • NPES - US Technical Advisory Group
  • ITU - International Telecommunications Union
    • ITU-R - Radiocommunication Sector
    • ITU-T - Telecommunication Standardization Sector
      • SG16 - Study Group 16 - Multimedia services and systems
  • JTC1 - IEC/ISO Joint Technical Committee 1
    • SC24 - Computer Graphics and Image Processing
    • SC28 - Office Equipment
    • SC29 - JPEG, MPEG, etc.


Industry Consortia & Associations

edit
  • AIC - International Colour Association
  • ANSI - American National Standards Institute
  • APCL - Association of Professional Color Labs
  • ASTM - American Society for Testing and Materials
  • ATSC - Advanced Television Systems Committee
  • CGATS - Committee for Graphic Arts Technologies Standards
  • DIG - Digital Imaging Group (Home of FlashPix and IIP)
  • EBU - European Broadcasting Union
  • EIA - Electronic Industries Association
  • FOGRA - Forschungsgesellschaft Druck (German graphic arts research institute)
  • GATF - Graphic Arts Technical Foundation
  • ICC - International Color Consortium
  • ISCC - Inter-Society Color Council
  • IETF - The Internet Engineering Task Force
  • IFRA - The International Association for Newspaper and Media Technology
  • JEIDA - Japan Electronic Industry Development Association
  • JBMA
  • MITI
  • NIST - National Institute of Standards and Technology
  • NPES - The Association for Suppliers of Printing and Publishing Technologies
  • NTA - The National Technology Alliance
  • NIDL - The National Information Display Laboratory
  • PIMA (US TAG for ISO TC42)
  • PIRA - The Information and Publishing Group
  • SIGGRAPH - ACM Special Interest Group on Computer Graphics
  • SMPTE - Society of Motion Picture and Television Engineers
  • SWOP - Specifications Web Offset Publications
  • TWAIN - Technology (or Toolkit) Without An Interesting Name
  • UGRA - Verein zur Förderung wissenschaftlicher Untersuchungen in der grafischen Industrie
  • VESA - Video Electronics Standards Association
  • VRML - Virtual Reality Meta Language Consortium
  • W3C - World Wide Web Consortium
edit

References

edit
  1. Shuttermuse glossary: image-quality = What does image quality mean when talking about photography?
  2. developer.nvidia gpugems3: importance-being-linear
  3. Shuttermuse glossary: image-quality
  4. Eizo: Ins and outs of HDR
  5. FractalForums.org: Tips for making high quality images
  6. Linux and High Dynamic Range (HDR) Display (By Andy Ritger, 2016)
  7. colorwiki: How To Get My Monitor To Match My Printer
  8. Slideshare: Digital-graphics-technology (By Becky J. Ritchie)
  9. howtogeek : what-is-pixel-density-and-how-does-it-affect-image-quality
  10. marketing-partners: Image resolution and quality what you need to know
  11. DFStudio: Digital image size and resolution, what do you need to know?
  12. Parker photographic: GIMP bit depth complete guide (2021)
  13. Petapixel 8, 12, 14 vs 16-Bit Depth: What Do You Really Need? (By Greg Benz, September 19, 2018)
  14. NineDegreesBelow: Users guide to high bit depth GIMP (By Elle Stone)
  15. Ted Lansing: 10 bit colour depth and Adobe Photoshop CS6
  16. graphicdesign stackexchange question: what-kind-of-black-should-i-use-when-designing-for-cmyk-print
  17. What is HDR and how can I get it on PC? Our HDR guide has it all Katharine Castle avatar Feature by Katharine Castle Editor-in-chief Published on July 9, 2018
  18. Bit depth (By Michael Reichmann, December 2, 2011)
  19. Eizo: Pixel density 4K
  20. Eizo: DisplayPort to D-sub
  21. Eizo: modern-video-inputs
  22. Computer Darkroom: Colour management (By Ion Lyons)
  23. Eizo: Colour Management System (CMS)
  24. Itsfoss: 6 Best Free and Open-Source Tools for RAW Image Processing in Linux (By Ambarish Kumar, February 4, 2021)
  25. BetterTechtips: 7 Best apps to edit RAW photos in Linux (By Aliko Sunawang, September 23, 2018)
  26. Docs GIMP 2.10: GIMP-image-precision
  27. Bit depth colour precision in raster images (By Denis Kozlov)
  28. color rendering intent by UNIVERSITY OF HOUSTON