What is the absolute relevance of bitmap editors?

[Xhris]

I want a discussion:

What can bitmap editing software (like Photoshop) do that vector software (in principle) can’t?

(This question should not be read as “what can Photoshop currently do that vector software can’t currently do”, or “please discuss the obvious distinction between bitmaps and vectors”, or “what would you change about Xara to stop it becoming a bitmap program”, and nor should this spiral into a Xara wish list or go off topic. Sorry for the strictness, but I want to say right from the off that I want it to be focussed and on topic and independent of Xara – Xara will not be mentioned again. I'm hoping for an earnest discussion.)

I’ve never really used bitmap editing software, and I’ve been flicking around on YouTube and the web etc. to see what typical uses people have for bitmap editing. I typically see photo-retouching, and changing the colour of objects such as hair etc. I see selection methods used, as well as warping/moulding of a selection, and the ‘healing’ of images. In principle, I can see all these functions being possible in a vector package, and all independently of resolution. I notice that the typical use of these editors is very object-oriented anyway through use of layers etc, but just being done at a fixed resolution.

Being more specific about the above uses I saw being used and how they could be applied (in principle) in a vector package: firstly, perhaps the most significant difference is the method of selection; a variety of tools/methods are employed to select regions of similar colour in bitmap editors. These range from magic wands, and other similar smart selection methods, to more sophisticated methods used when selecting objects with complex edges – such as fur. These methods could conceivably be applied to bitmaps and/or a collection of vector objects while in a vector package to select regions of similar colour.

Secondly, once regions are selected/masked; changing the colour properties (hue/sat/val etc) of the selection is done using a number of tools. Again all this could be done in a vector package – and to an extent, already is. Warping/moulding of selections using a variety of tools could be achieved in vector packages too; in the case of moulds: this is already capable to a certain extent. With tools like smudge etc, this again is also done to an extent in some vector packages (on vectors in this case – by pushing nodes around etc).

It recently occurred to me how ‘resolution independent’ operations could be performed on any bitmap and vector selections. A selection could be mapped to a high resolution matrix, and warping operations, for example, are simulated by manipulating colours on the vertices of the grid. This ultra-fine grid increases the effective resolution, allowing for such fancy features. In a vector package, any traditional “bitmap” tool could work in the same way; the selection is mapped to a high resolution matrix (with resolution determined by the internal maximum resolution of the product), and the transform operations are performed on that matrix (at the resolution being viewed at – i.e. the zoom factor/export resolution etc).

This would allow all effects to be produced in a vector manner; take ‘healing’ tools for example: the region to be healed is typically regenerated from an interpolation of the region surrounding the active region.

Objects would not even necessarily need to be selected to be operated on; imagine a collection of unselected objects in a vector package, and you dragged across them with a smudge/smear tool. It could apply a quasi-object over the drag region, and within the bounds of that quasi-object, apply all the effects desired given the information underneath using the method discussed above.

I like to think I see a future where only vector packages exist; I suspect that the reason why bitmap editors are still around, is that all these typical “bitmap” features are together already in one place within them. No one vector package has such tools that can be performed on both bitmap and/or vector groups as seen in these bitmap programs. In the long term, I just can’t see a future where there is a distinction between the two types of graphics package. I suspect (in an ideal world) future advanced vector products will incorporate all the tools (i.e. tool functionality) currently found bitmap editors, and become the dominant graphics-package-form.

But are there any (specifically) advanced tools/features/actions that are only possible to be performed in bitmap editors (i.e. to bitmaps only)? I can’t think of any given the method discussed above.

[jamesmc]

With the feathering tool, it's hard to think of anything that couldn't be done.

I would not include the plug-ins as I think most all are for bitmap effects.

I also would not include the brushes tool as I think it is a bitmap effect.

All one has to do is watch one of Bob's videos on cartoon drawing, to understand why vector can be easier and more powerful than bitmaps.

For the accomplished artist (no offense to any meant) a bitmap program may be what they are used to doing. You would have to ask the Corel Paint folks about that.

Where bitmap programs come into play imo, is that when you want to do some pre-packaged effects that most vector editors do not have even in plug-ins.

These things include effects like pallet knife, oil brush and other effects on the ready, that one would have to make from scratch in a vector editor.

So, what it boils down to is imo, technique, skill and convenience.

It takes different skills to be a vector artist than it does a bitmap artist.

I use Xara Xtreme 3.2 and Corel PSP XI. That's about all the power I can handle with my current level of artistry (cough.)

[David O'Neil]

What can bitmap editing software (like Photoshop) do that vector software (in principle) can’t?

Select all of the pixels in a drawing of a certain color value, with or without a tolerance, and change all of them in one fell swoop to another color.

I did not spend much time reading your essay, but it seems to me that if you are going to make a matrix out of a bitmap, you really have a bitmap under another name, so you would have a bitmap editor in the end, that claims to be a vector editor, but is not really. Anyway, Xara's bitmap tracer almost gives you the 'matrix' you are talking about. So now you have ten million 'objects' rather than ten million 'pixels' when your trace gets accurate enough. Is that really what you want? What has it gained? When you have to magnify the drawing 1000 times to place your drawing on a wall rather than a sheet of paper, all of a sudden you have huge square 'blocks' in your printed output which used to be individual pixels in the original bitmap.

I might be wrong, as I might not understand your concept correctly.

[steve.ledger]

I want a discussion:

What can bitmap editing software (like Photoshop) do that vector software (in principle) can’t?

What's the next Q? Should digital cameras capture as vector based images?
Come on - horses for courses Xhris.

[handrawn]

Scanning?

They are very different techniques raster and vector - love james quote from another thread:

Me? I'm a hammer, saw and ruler type of guy - vector graphics suit me. Some artists are a pallet knife, brush, chalk, pencil, spray paint, masking kind of person.

I use both programs and wouldn't ever change it.

Exactly.

Oh and anyone about to give up filling shapes with bitmap textures etc anytime soon ?

[Grafixman]

Speed and simplicity perhaps?
You could emulate most bitmap editing methods with vectors, but it's a more involved process than when done with a bitmap editor, at least for me. And if you're up against a deadline, which method would you choose?
Also, bitmap brushes. I don't think vector brushes have the ability to pick up and smudge underlying colors. Even smoothing and smudging. You could emulate them if you're patient enough, but it's just not the same thing.

[Xhris]

(All my responses to debates are typically strident, but said in the nicest, loving tone. )

The comments so far are of the type I was fearfully expecting, but tried hard to avoid when starting this discussion.

Select all of the pixels in a drawing of a certain color value, with or without a tolerance, and change all of them in one fell swoop to another color.

I did not spend much time reading your essay...

Oh dear...you can stop right there... I don't think I could ever comment on something I hadn't even read.

All vector graphics are rasterised at some point. E.g. at display or export time.

Come on - horses for courses Xhris.

This line of thinking is actually not so. "Electricity and magnetism...the same thing? Go burn some witches!" "space curvature?! get this guy a straight jacket" "small particles are waves?! no PhD for you de Broglie, you nutcase".

With a bit of thought, it's fairly easy to see there need be no necessary distinction between bitmap and vector operations. And the idea of vector capture by cameras (i.e. a 'smart' tracer) I came up with a long time ago. It would require stupendous levels of research to emulate AI recognition techniques, but is easily conceivable.

Anyway, back to the question: What could bitmap editing software (like Photoshop) do that vector software (in principle) couldn’t?

Speed and simplicity perhaps?
You could emulate most bitmap editing methods with vectors, but it's a more involved process than when done with a bitmap editor, at least for me. And if you're up against a deadline, which method would you choose?

This perhaps goes back to the point I was making about the typical bitmap tools not currently being in a vector package. Understandably, given how the two packages evolved and why, but I see convergence eventually - which is essentially what this discussion is supposed to be about.

[David O'Neil]

Oh dear...you can stop right there... I don't think I could ever comment on something I hadn't even read.

All vector graphics are rasterised at some point. E.g. at display or export time.

I never said I hadn't read it--I had. It just wasn't very clear, and I wasn't going to spend another hour rereading your essay.

From the understanding I got, it seems that your 'rasterization' process is not truly vector--it is bitmap in disguise. I pointed out that when scaled, you don't have the same thing as a true vector drawing, and you will therefore see artifacts of that process.

[humorous tone] Is that clear now, Mr. Loving Snip? [/humorous tone]

[Xhris]

...when scaled, you don't have the same thing as a true vector drawing, and you will therefore see artifacts of that process.

This assumes that the effect isn't reprocessed (like all vector effects are - e.g. feathering) when the object is scaled, or effectively-scaled in the case of a change in zoom factor. Therefore, it is possible to have vector-like bitmap effects. This is the consciousness-raising issue I'm highlighting, the distinction is arbitrary between the two graphics types - one is just effectively resolution independent. An understanding of what vector means in graphics software maybe needs highlighting too. Mathematically, vectors represent direction and magnitude, but when displayed on screen, a representation of that is presented in a fixed-resolution, pixel form. This is recalculated for any change in the perceived resolution. In effect, all vector output is bitmap. With vector software however, the resolution can be changed to suit.

[David O'Neil]

Find a 100x100 pixel thumbnail anywhere on the net. Now, using any pixel editor, increase that to a 5000x5000 graphic without doing any touch-up painting. Even using smoothing algorithms, the big version will never look as good as if you increased a true vector illustration's output size from 100x100 to 5000x5000. The two types of drawings are inherently different. (The 'outputs' are not different, the 'types' are different.) That is what you are overlooking. Vectorizing the bitmap will only enable you to get as good of output as the smoothing algorithms that already exist for bitmaps.