TGJ3M Raster, Vector, Resolution and File Formats: Notes and Practice

Key things to remember

1Raster images are grids of pixels and degrade when enlarged; vector images are equations and scale losslessly.

2Logos and icons must be built as vector — a flattened raster logo can never be made sharper.

print size = \frac{pixel dimension}{PPI},

and

pixels = inches \times PPI

4300 PPI is the standard for professional print; 72–96 PPI is the screen convention.

5A colour depth of

n

bits gives

2^{n}

colours. 24-bit is 8 bits each for red, green and blue — about 16.7 million.

6Uncompressed size in bits is width × height × bit depth. Divide by 8 for bytes.

7Lossy compression (JPEG) discards data permanently and degrades on every re-save; lossless (PNG) does not.

Every image you will ever make is one of two things, and almost every technical problem in graphic design comes from confusing them. A raster image is a grid of coloured pixels. A vector image is a set of mathematical instructions for drawing shapes. That single difference determines whether a logo survives being blown up onto the side of a bus.

Fig 1Enlarge a raster and you enlarge its pixels. Enlarge a vector and it is simply redrawn.

ARaster (bitmap)

A fixed grid of pixels
Enlarging beyond its native size produces visible blocks and blur
Handles photographs and continuous tone
File size grows with dimensions and colour depth
JPEG, PNG, GIF, TIFF, PSD

BVector

Paths, points and fills described by equations
Scales to any size with no loss at all
Handles logos, type, icons and flat illustration
File size depends on complexity, not dimensions
SVG, AI, EPS, PDF

The rule that follows from this

Build logos, icons and anything that will be reproduced at many sizes as vector. Once a logo has been flattened to a raster, that resolution is the best it will ever be — you cannot recover detail that was never recorded. 'Just increase the resolution' does not work, because upsampling invents pixels rather than finding them.

Resolution: pixels, PPI and print size

Fig 2The same pixel count printed at three densities. More PPI means a smaller, sharper print.

An image's pixel dimensions are its real content — say $3000 \times 2000$ pixels. PPI (pixels per inch) is not stored in the pixels; it is an instruction about how densely to lay them down when printing. Change the PPI without resampling and the pixel count is untouched, only the printed size changes.

print size (inches) = \frac{pixel dimension}{PPI}

Rearrange it to find the pixels you need: pixels = inches × PPI.

Destination	Resolution to use	Why
Screen / web	72–96 PPI (or just pixel dimensions)	Screens address pixels directly; PPI is meaningless to them
Desktop / office print	150 PPI	Acceptable on plain paper
Professional print	300 PPI	The industry standard for photographs in print
Large-format banner	50–150 PPI	Viewed from metres away, so fewer pixels per inch are needed

Downsampling is safe; upsampling is not

Throwing pixels away to make an image smaller loses detail you no longer need. Adding pixels to make it bigger means the software is guessing at colours between the originals, which produces softness rather than detail. Always start from the largest original you have.

Colour depth and file size

Fig 3Each extra bit doubles the number of available colours.

Colour depth (or bit depth) is the number of bits used to record each pixel's colour. With $n$ bits you can distinguish $2^{n}$ different values, so 8-bit gives 256 and 24-bit gives $2^{24} \approx 16.7$ million — the '24-bit true colour' of 8 bits per channel across red, green and blue. 32-bit is the same with an 8-bit alpha channel added for transparency.

file size (bits) = width \times height \times bit depth

Divide by 8 for bytes, then by 1024 twice for megabytes. This is the uncompressed size.

This is the uncompressed size

The formula gives the raw size. Real files are almost always compressed, so a JPEG will come out far smaller. The calculation still matters: it tells you what the image costs before compression, which is what determines memory use while editing.

Compression, and choosing a format

ALossy compression

Permanently discards data the eye is least likely to miss
Much smaller files
Quality degrades every time the file is re-saved
JPEG, MP3, most video codecs

BLossless compression

Rebuilds the original exactly
Larger files
Re-saving costs nothing in quality
PNG, GIF, ZIP, TIFF (LZW)

Fig 4Work backwards from the destination and the answer is usually forced.

Format	Type	Transparency	Use it for
JPEG	Raster, lossy	No	Photographs for web and print where file size matters
PNG	Raster, lossless	Yes (alpha)	Screenshots, flat graphics, anything needing transparency
GIF	Raster, lossless, 256 colours	Yes (1-bit)	Simple animation; obsolete for still images
SVG	Vector	Yes	Logos and icons on the web — scales to any size
TIFF	Raster, lossless	Yes	High-quality print masters and archives
PDF	Both	Yes	Sending a finished layout to a printer or a client
PSD / AI	Working files	Yes	Editable masters with layers — never a delivery format

Keep the master

Always keep the layered working file (PSD, AI, or the project file) and export flattened copies from it. An exported JPEG cannot be un-flattened, and the client who asks for one word changed will ask on the afternoon of the deadline.

Worked examples

Reveal one step at a time and try to beat it to the next line.

Example 1

A photo is

3000 \times 2400

pixels. How large can it be printed at 300 PPI, and what size would the same file make at 150 PPI?

Example 2

A poster must print at 24 × 36 inches at 150 PPI. What pixel dimensions does the image need?

Example 3

Calculate the uncompressed file size of a

1920 \times 1080

image at 24-bit colour depth, in megabytes.

Example 4

You need to put a school logo on a website, and the same logo will later go on a banner. Which format(s), and why?

Where marks get lost

Trying to rescue a small raster logo by raising the PPI. Upsampling invents pixels; it cannot recover detail that was never captured.

Treating PPI as a property of the file. Pixel dimensions are the content; PPI is only an instruction about print density.

Saving a JPEG repeatedly while editing. Each save re-applies lossy compression, and the damage accumulates.

Using JPEG for a logo or a screenshot, which produces visible artifacts around hard edges and offers no transparency.

Dividing bits by 1000 instead of 8 to get bytes. Bits to bytes is ÷8; bytes to kilobytes is ÷1024.

Delivering a flattened export as the master and keeping no layered file, so the smallest revision means starting again.

Designing a large-format banner at 300 PPI, producing an enormous file for something viewed from several metres away.

Key things to remember

Worked examples

Where marks get lost

Flashcards 11

Practice