Graphic design tools and what each one is actually for
The deliverable picks your graphic design tool: anything that will be scaled up or printed needs vector, anything photographic needs pixels, and interface design has a tool of its own. There are six categories, and in each of them at least one option costs nothing and still produces professional output.
- Lesson 8 of 12
- Beginner
- Free, no signup
Six tool categories, around the file you have to hand over
Each category produces one kind of output. Tools inside a category compete; tools from two different categories do not replace each other.
One application can cover the first three; the interface category and the AI category genuinely need their own tools.
Last checked: Facts and tool names in this lesson are re-checked against their sources on this date.
Why the right question is not the name of an app
The beginner question is usually: should I learn Photoshop or Illustrator. The answer is not in either name. It is in where the file you hand over is going to end up.
Picture a logo delivered as a PNG 1024 pixels wide. On the website it looks right and everybody is happy. Six months later the same client wants a shop sign and that file is worth nothing, because a pixel image has a size and anything past that size is just blur. Had the mark been vector, the sign, the business card and the site icon would all have come out of one file.
So the first dividing line between tools is not which is more expensive or more popular. It is whether the output is pixels or vector. Photos, textures and retouching are pixels. Marks, icons, lettering, maps, anything whose size changes later, are vector.
After those two there are three more categories, and none of them substitutes for another: multi-page layout for print, interface design which works in components, states and prototypes, and template tools that sell speed and take freedom in exchange. The sixth category is new: AI image generation, which gives you raw material rather than a finished file.
Our position is blunt. The software itself can be learned in a week and they all resemble each other anyway: layers, masks, paths, type. What you do not learn in a week is telling which deliverable needs vector and which file has to still open in five years. That judgement is what separates professional work from amateur work, not the list of applications you have installed.
Will this output be scaled up later, or printed?
A vector tool
- Logo and mark
- Signs, banners, billboards
- Icons and icon sets
- Diagrams and maps
A pixel tool
- Retouching and photo correction
- A social post image
- Texture and background
- Compositing several photographs
A vector file with a photograph placed inside it is no longer all vector; that photograph still breaks when you scale it up.
Six tool categories and the job each one does
Each category answers one question. Tools inside a category compete with each other; tools from two different categories do not, however much their marketing would like them to.
| Category | The question it answers | Usual output | Example tools |
|---|---|---|---|
| Pixel | How do I fix this photograph | PSD, TIFF, PNG, JPG | Photoshop, GIMP, Affinity Pixel studio |
| Vector | How do I build a mark that takes any size | SVG, PDF, AI, EPS | Illustrator, Inkscape, Affinity Vector studio |
| Layout and print | How do I set several pages for a printer | Print-ready PDF | InDesign, Affinity Layout studio |
| Interface | How do a product's screens work together | Cloud file, SVG and PNG exports | Figma |
| Templates | How do I make one post or banner quickly | PNG, JPG, PDF | Canva |
| AI generation | Where do I get raw visual material | PNG of variable quality | Image models |
One application can cover the first three today. Affinity's own page says it brings vector, photo and layout tools together in one place. The fourth category is genuinely separate though: interface design deals in components, states and prototypes, and no photo editor has those concepts at all.
The sixth category needs one sentence. What a model produces is raw material. For sketching, mood boards and background texture it is excellent. For a logo it is not, because you have to hand over a vector file and you have to be able to say where the mark came from. Our ranking of image models, with criteria and sources, is in AI image generation, and the Iran access column lives there too.
How far do the free ones actually go?
In the first three categories the free option is not a compromise. GIMP describes itself on its own page as "a freely distributed program" for photo retouching, image composition and image authoring, developed as community free software under the GNU GPL. Inkscape is a free and open source vector editor under the same licence, for GNU/Linux, Windows and macOS, and more to the point its main format is SVG itself; its own page lists SVG, AI, EPS, PDF, PS and PNG among the formats it imports and exports.
Affinity went further than that. Its own page says: "Yes, Affinity really is free. You can use every tool in the Pixel, Vector, and Layout studios, plus all of the customization and export features, as much as you want, with no restrictions or payment needed." It is available for Mac and Windows, with iPad listed as coming soon. Two things are stated plainly there as well: you sign up to download it, and Canva's AI tools inside Affinity unlock only on a Canva premium plan.
Figma is a different story, because free there means limited rather than stripped down. Its pricing page gives the Starter plan "unlimited drafts", UI kits and templates, and "150 AI credits/day, up to 500 AI credits/mo", while its help page states the real limit: a single team with one folder and 3 files, and 30 days of version history. The Professional plan is listed at $16 a month per full seat.
About Canva this lesson prints no numbers. Their own pricing page did not open from our server, and we do not publish second-hand figures about anybody's free plan; it is the same rule we keep across this site.
The real cost of the free route sits somewhere else and nobody dwells on it: file exchange. A client or a printer who sends you a .ai, .psd or .indd wants the same thing back. Opening those files in another tool is usually possible, but "opens" and "looks identical" are not the same sentence: effects, fonts and some print-specific features are what get left behind in the conversion.
What is actually available from Iran?
Here we write only what the companies' own documents say, and we link them. The rest we do not know, and we say that we do not know.
Adobe, in its own end user licence, writes that the export and re-export of its software products are controlled by the United States Export Administration Regulations and that such software may not be exported or re-exported to Cuba, Iran, Iraq, Libya, North Korea, Sudan, Syria, or any country to which the United States embargoes goods. That is Adobe's own sentence, in Adobe's own document, and Iran is named in that list.
Figma has its own export control clause in its terms of service: use of the platform is subject to export control and economic sanctions laws including the US Export Administration Regulations and the rules administered by the Office of Foreign Assets Control of the US Treasury, and Figma writes that it may take measures required to comply, such as suspending access to the platform. Again: their words, not our reading of them.
Let us be equally plain about what we do not know. We do not test these services from Iran and we make no claim about what happens on your connection today. We also found no page from Affinity publishing a country list, so Affinity's status for Iran is unknown in this lesson and stays unknown. If a service's page does not name your country, that means unknown, not yes.
And one thing we do not write here: how to get around any of it. Not because a route could not be described, but because this is a legal and contractual question and answering it on your behalf is not our job. What can be done instead is this: free software has no purchase and no account step at all, and you can build your working chain on formats that depend on nobody's account or subscription.
Which file stays loyal to you
Tied to nobody's account
- SVG for anything vector
- PDF for print and archive
- PNG and TIFF for pixel images
- A plain-text note of fonts and colour codes
Depends on an account or a subscription
- A file that only lives inside one cloud service
- A source format only one application reads fully
- A font you do not hold a licence for
- An effect that only opens with one plugin
An open format means the file opens, not that it looks identical; effects and fonts do not always survive the trip.
A toolchain you can actually work with
Four recurring jobs, and the route each takes from tool to handover.
- Logo. Sketch on paper or take raw material from an image model, then redraw it in a vector tool. Handover: the vector file as the source, plus SVG for the web, PDF for the printer, a few transparent PNGs at the sizes needed, and a single-colour version.
- Social post. A template or your own fixed-size file, exported as PNG or JPG. Speed matters more than flexibility here and a template-driven tool is doing the right thing.
- Print catalogue. A layout tool, CMYK colour, images at a resolution fit for print, and a print-ready PDF with the settings the printer asks for. Get the trim size and marks from the printer rather than guessing them.
- Website or app interface. An interface tool, components and styles from the first hour, SVG exports for icons. What a developer needs is not a beautiful file; it is names and values.
One habit that keeps the work clean: put a short note next to every handover listing the font names, the colour mode and the primary colour codes. It is three lines and it prevents a rebuild six months later.
And the sentence this whole lesson stands on: what you hand over is a file, not an application. The client has no interest in the name of your software; the client has an interest in the file opening and printing correctly.
The fast path, with AI
The fast path here is not asking a model which application to install. It is turning your list of deliverables into a spec table before you open any application: every item with its tool category, source format, export formats, exact size, colour mode, and a "this must be vector" flag. That table is what prevents the rebuild six months later. A cheap fast model is enough, and our current pick is in the <a class="text-link" href="/en/ai/">AI section</a>.
- Write every single thing the client actually asked for on one line, and say where it will be used. "Logo for the website and the shop sign" is a line; "stationery set" is not, because nobody knows what it contains.
- Hand the list to the model with the prompt below. The output must be a table whose last column is "question": anything the model could not derive from your list has to become a question, not a guess.
- Send the question column straight to the client. This is the one step people skip and then come back to in the middle of the work.
- Review the flagged rows yourself. The model does not know whether that sign is three metres or thirty centimetres; you do, and that one column decides whether the source file has to be vector.
Copy-ready recipe
Role: production manager at a design studio. You have only the list below.
Deliverables list:
{one item per line, with where it will be used}
Build a table, one row per item, with these columns:
1. Deliverable
2. Tool category (pixel / vector / layout / interface / template), not a brand name
3. Source file format
4. Export formats that must be handed over
5. Size with its unit
6. Colour mode (RGB or CMYK)
7. Must it be vector? Yes / No, and why in one sentence
8. Question: anything the row needed that was not in the list
Rules:
- Name no software. Column two is a category only.
- Do not guess a size that does not come from my list. If you use a
platform's standard size, mark it (suggested) so I can check it.
- Any item that will be scaled up or printed gets Yes in column seven.
- At the end, list the questions separately and consecutively, so I can
send that block straight to the client.
- Add nothing to the list. If an item is needed and I did not name it,
put it in the questions, not in the table.Before you trust the output: Two columns are not the model's to decide and are yours: the real size and the colour mode. Get the size from the client or the printer, and the colour mode from where the file is going. For every row that came back Yes in column seven, make sure before you start that your source file really is vector and has no photograph placed inside it.
AI in this kind of work
In tool selection AI is not a category; it is a layer sitting on everybody else's work. What it does well is organising a client's wishes into a technical spec. What it does badly, and what everyone asks of it, is telling you which application is better, or attributing a feature to an application that does not have it.
Tools that actually help
- Gemini Suits turning a messy brief into a spec table, because it is comfortable with long multi-item text. Our own entry notes Iran is not on Google's supported-countries list; the payment route is in <a class="text-link" href="/en/ai/buy/">buying AI access</a>.
- Claude Better when you want the client's own sentences preserved verbatim in the questions column, because it paraphrases less. Iran is on neither of Anthropic's two supported-countries lists.
- RGB AI image ranking Not a tool itself, and deliberately here: six image models ranked with criteria and sources, including an Iran access column, so instead of asking "which model is better" you can read the table.
Where it backfires
The specific risk in this subject is invented features and invented sizes. Ask a model which application does a certain thing and it will build a list in which some rows do not have that capability or refer to a version that never existed; because the question looks like questions that have answers, the answer comes out looking like an answer. The same happens with sizes: platform standards change and the model states an old one with complete confidence. That is exactly why every size in the prompt above that did not come from your list has to carry a suggested tag. The second risk is about data and belongs to your client: a brief for an unreleased project carries the client's name and budget; sending it to an outside service is a decision, and whether the conversation is used to train a model depends on that service's plan and settings and is written on its own page. And a third risk specific to this craft: an image a model produced is not a vector file, and whether you may use it commercially depends on that service's terms. For a mark that is going to be registered, those two together mean the model's output is a starting point rather than a handover.
Sources: Anthropic: is my data used for model training Google: Gemini Apps Privacy Hub
Where this advice stops
This lesson does not tell you which service comes up on your connection today; we do not test that and we make no claim about it. Adobe's sentence and Figma's clause were read from their own documents on this page's verified date, and documents change. Most of all: no tool fixes weak design. Somebody who understands composition, spacing and hierarchy produces better output in GIMP than a beginner does with the most complete paid suite.
From our own work
The page you are reading is itself an example of a tool decision. Every infographic on this site, including the hexagon at the top of this lesson, is drawn with HTML and CSS rather than in a design tool: our shape library holds 52 shapes and its files contain not a single img tag. What that bought: the export step disappeared, no double-resolution copies were needed for retina screens, and because the text inside a shape is genuinely text, the same figure reads correctly in four languages and in both reading directions. What it sold is equally clear: not every shape comes out this way, and we only have the shapes we have written code for.
Real follow-up questions
Should I learn Photoshop first or Illustrator?
It depends on the work you want. If your work is logos, icons and print, the vector tool comes first; if it is photos, retouching and social posts, the pixel tool does. If you do not know the answer yet, learn vector first, because a vector file can always be turned into pixels and the reverse is not true.
Can you deliver professional work with free tools?
In terms of output, yes. The client sees a file and the file does not say what made it. Where it gets hard is exchanging source files with people who work in the paid suite and expect those formats back.
Which format should I hand the final file to the client in?
One source file plus a few exports. For vector work: the source, plus SVG, PDF and a few transparent PNGs. For pixel work: the layered source, plus PNG or JPG at the sizes needed. Next to it write three lines naming the fonts and the colour codes.