My name is Paul R. and I am a visual artist living in The Netherlands.
I am new to do software and want to know a couple of things.
1. Has anyone done larger prints for gallery work ? What are the challenges when it comes
high res print (larger then 7500X5500 300 dpi), colors, is the output RGB ? Is everything still razor sharp. Of course any examples will be welcome.
2. Is there a way to add several images. So they become all sources and layers ? And can you work on them as a layer. Or will the next images be the new source and is the old one just gone ?
Lot's of people have done large prints from Studio Artist over the years. You sometimes see them in art galleries for tourists in Honolulu sometimes presented as 'real' art work (whatever that means).
If you filter using the exhibition option on Studio Artist News, there are a bunch of posts there.
And we'd certainly like to include more posts there, so if people are doing Studio Artist generated exhibitions let us know about it.
There are a number of different approaches to generating output, so i'll discuss them a little.
Since you are a new user, i'll also include some historical perspective.
Re-rendering the canvas
When Studio Artist was first released, one very unique feature was the dual raster-vector graphics model we used. Raster paint was applied to vector paths.
And you could record what you were doing (possibly in a lower resolution working canvas) as a History Sequence or Paint ActionSequence, then re-render that PASeq to a higher resolution canvas.
Here's a tip that discusses the whole re-render concept.
Re-rendering is more like making an edition print. Because the nature of how randomization works within Studio Artist, combined with what is often described as 'the butterfly effect' in chaos theory, means that while the re-rendered painting looks similar to what you originally did, there are random variations within it.
But of course Studio Artist grew in feature capabilities and complexity over the years. And as the paint synthesizer became more elaborate, the original raster paint nibs applied to a vector path model started to be perverted or challenged by some of the newer features.
For example, in most paint programs, a brush is an image being dragged around on the canvas. You can certainly do that in Studio Artist. But what a user perceives as a brush might actually be composed of a number of much smaller paint particles that might have very sophisticated internal behavior.
Later, vector paint effects were added to the paint synthesizer. These very quickly were additionally perverted by newer features also added to the paint synthesizer.
A good example of what i mean by this would be 'region fill as brush' pen mode.
The original paint synthesizer model was raster paint nibs applied to a vector path. But in 'region fill as brush' pen mode, the vector path is used to create a one time paint brush and associated paint nib on the fly.
So rather than applying raster paint nibs (or vector paint nibs) along a vector path, the vector path is actually used to create the paint nib (which again could be raster or vector).
So what is the brush size of a 'region fill as brush' pen mode paint?
Well that's a very interesting question, and it can be very complex, since it depends on the spatial extent of the particular path that is being automatically (or manually if a user is hand drawing it) drawn, and then converted into a one time brush that is sued to create the nib of paint that is applied to the canvas.
So in addition to educating you a little bit about the underlying complexity of the paint synthesizer, i'm also trying to point out that our original re-render the paint model to blow it up may or may not work well, depending on which paint synthesizer features you are using.
Vector graphics output
Anything associated with vector paint or other vector drawing processes (like the Vectorizer) can be output as a vector graphics file if you so desire. We offer EPS andSVG vector output. But at this point in time, the SVG vector file output is the most fully developed. This is because it directly matches up with our internal vector graphics engine.
EPS pre-dates SVG support in Studio Artist. We even had eps output as an option in SA V1 back at the dawn of time before we even offered vector paint options. There were various ways to output just the vector path, or to kludge a vector something that tried to track some aspect of the paint, so we could output eps for people to monkey with in other programs.
But EPS does not directly support partial transparency. SVG does. So EPS output may not look exactly like what is on the canvas, since most canvas painting makes heavy use of partial transparency, and that is not going to be reproduced in the SVG output.
So again, SVG is really the way to go if you want to go this route. Here's a tip on SVG output.
If you go this route, you need to make sure what you are working with is really a vector paint or vector effect. It's easy to configure the Vectorizer to output vector. It's certainly easy to configure the paint synthesizer to output vector paint, but it's also possible to configure it to create vector-raster hybrid paints. Many of the 'wet look' vector paints are really vector-raster hybrids. An if you do that, then only the vector part of the painting process is going to output as SVG.
I should point out that we also have full vector wet paint options available in the paint synthesizer. And i believe we are unique in that regard. Pushing the envelope of what is possible.
It's possible to configure paint presets in the paint synthesizer that appear to be raster natural media paint, since they have the visual complexity of that, but in reality they are full vector paint. So they can be output to SVG.
Since life is usually more complicated then you might first suspect, that also applies to SVG output of these natural media look paints. Because the SVG file format is not really designed for what we are doing, so it is very easy to generate huge SVG vector files. So big they will crash Illustrator as it struggles to open them. usually you can still open those in a web browser, so it's a limitation of Illustrator, not a problem with the files themselves.
Of course Studio Artist does not have any problems opening these huge SVG files. So you can always import them in Studio Artist to the output canvas size you want for your final print.
Be aware that it currently is possible to create SVG files of such complexity that Studio Artist crashes when trying to output them. We know what the problem is. It's not in our code, it's in the application framework code we currently use. They keep the entire SVG file in memory, as opposed to spooling parts of it to disk, before combining them all together for the final spool out to a SVG file, and they store them in a kind of string variable that cannot be indefinitely resized in virtual memory. Our plans are to address this in some way in the future so it's not a problem.
When working this way, it usually ( i would say always) makes more sense to do your artist creation at a lower resolution than what you want to use for your final output.
Now i realize this is very alien to your typical graphics artist used to working in programs like Photoshop.
And it takes a while to wrap your head around it as a working concept.
Studio Artist is more than just a paint program
The Paint Synthesizer is just one art of Studio Artist. In addition to the Vectorizer, there are also Image Operation effects, MSG (modular synthesized graphics), Interactive Warp and Adjusts, Temporal Image Operation Effects, etc. These can all be combined together into paint Action Sequences.
So my discussion of resizing paint presets. And my discussion of vector output don't necessarily apply to the other operation mode effects you might be using inside of a single PASeq preset.
Thanks for the answers to my questions. Quite a bit of information :) and not always easy deal with it :).
For me it's still pretty much a challenge to come to high resolution ready to print images when exporting it from SA. If I put in a 7500 x 5000 pixel image at 300dpi, I sofar can't export out of SA the same quality. I am on mac 64 bit. all the images export as 72 dpi, even if you do the resize or change it to 300 dpi before you export. I do understand that you can change the 72dpi into 300 dpi in Photoshop without resampling but then your print gets smaller. So in other words you will always lose a significant amouny of quality.
For using it online, etc. I agree it doesn't matter.
Why is the program not keeping it to 300 dpi when you export ?
Hope this so impressive software will give me the right needed print quality...
You don't lose any quality. pixels are pixels. The dpi is meta data laid on top of the pixels in the raster image. And yes, that dpi meta data is used when computing the physical size.
But there is no quality lose when outputting at 72 as opposed to 300 dpi. because if you change the dpi metadata in the other program to be 300 (without doing any resampling of the pixel data in that second program), you then would have what you would get if the 300 dpi meta data did tag along.
Not all of the image writers we support implement the dpi meta data write. So that's why the issue is there.
That said, I'm happy to take another look at what's going on to see if we can improve the behavior in V5.5. So thanks for the feedback.
Ok, so i made some changes to SA V5.5 to help you out. You should see the dpi metadata pass through now if the image write ends up going through our primary raster image write tier. There are multiple tiers in case certain functionality is missing on a user's computer. I'll run this down through the other tiers when i get the time.
Each layer in Studio Artist consists of a ARGB raster image, and a vector path frame composed of bezier paths.
When you save the canvas using File : Save Canvas as menu command, you are saving a raster image file. You can choose the file format for output.
The raster output is either RGB or ARGB depending on what file format you choose, and whether alpha vie is enabled or not for the canvas.
Unlike some other programs you may be familiar with (photoshop for example), alpha view can be turned on or off. It defaults to off. You can use the alpha channel of a layer for alpha masking, but you are free to use it as a temporary buffer for computations you are creating in a PASeq, so it could be used to store additional selection masks, images to modulate with, etc. Studio Artist is very configurable.
Her's a tip on working with alpha channels in layers if you want to learn more about this.
Programs like Photoshop encourage you to create a thousand different layers. We actively try to discourage you from taking this approach. many of the things you would need multiple layers to do in Photoshop you can do in a single layer in Studio Artist. All Studio Artist effects have Compositing built into them.
So say you want to add an additional black edging effect to something you painted. In Photoshop, you'd create a second layer to do that. In Studio Artist, you can just run the edging effects (let's assume it is a Vectorizer or Ip Op) with the Composite option for that effect set to Min.
Photoshop is also designed from it's very core to store as little of the canvas in memory as possible. it works on chunks of the image at a time.
Studio Artist stores the entire image in memory. By doing so, we are able to use some super cool image processing algorithms that you just can't really use when you take the other approach. The tradeoff is that Studio Artist is a memory hog. Because everything you are working on is in memory.
32 bit applications are limited to a theoretical limit of 4GB of virtual memory. No more (even is you computer has 32 GB of ram). One a mac you can get very close to that 4GB limit. On windows it's a different story. You are limited on windows for 32 bit apps to 2GB minus whatever memory the dynamic libraries currently being used on your computer are taking up). This limits you to about 5000 pixel canvas size before you start running into serious memory limitations.
And it depends on what you are doing as well. Many Studio Artist internal operations may work at a higher precision than the 8bit per channel, 32 bit per pixel raster frame buffer in a layer. Most Ip Op effects are actually computed using 16bit per channel internal buffers. The Vectorizer needs to use a double sized internal buffer during part of it's calculations.
One the mac, V5 also includes a second SA5 64 bit application. This gets you around the virtual memory limitations. On windows, V5 only has a 32 bit SA5 app.
We are currently working to finish up Studio Artist V5.5. This is 64 bit only on both the mac and windows. So the virtual memory limitations that our current V5 users experience on windows are gone in SA V5.5.
Some of our users have used external interpolation for upsizing their Studio Artist generated working canvases for final print size.
We have something inside of Studio Artist called SuperSizer. This can be useful in certain situations. I've created commercial compact disk artwork that was later printed from very small original image sizes using SuperSizer, so it definitely is useful in certain situations.
Other people import into Photoshop, and use either the build in PhotoShop interpolation, or various plug-in interpolation effects to get to the final size for print.
Fractal interpolation software has been popular with some users.
A new development in this area is neural network interpolation software.
We are actually very interested in neural net interpolation, and are actively researching that internally. So you can expect it to pop up in the Studio Artist universe in the future.
Which bring me to another related topic, working canvas sizes.
Often, the best working canvas size is smaller than your final output print size.
And we are very focused on trying to create an eco-system to give you options for working this way, while still getting the quality you want in your final output print size.
I've discussed re-rendering a recored History Sequence. I've discussed vector graphics file output options. And i've mentioned 'smart' interpolation, like neural net or fractal interpolation.
Let's just consider using the Vectorizer for a moment. Using a smaller working canvas size is typically going to look much better for Vectorizer output than using a huge working canvas size tied to your final desired output canvas. At least i have found that to be true. And in discussing it with other artist, they come to a similar viewpoint.
I should point out that some Ip OP an MSG effects are also generating vector information internally, and these can be output as SVG vector graphics as well. These 2 areas are kind of frontier territory for SVG output right now, we're working to make them easier to use in future updates.
And any effect is going to be much more responsive at lower working canvas sizes.
We are very committed as a company to developing workflow options that will enable artists to build their canvases at working canvas sizes that could be smaller than their desired final output size, and then giving them the options they need to get to that final output size with the quality they desire.
So i think there are ways to get to where you want to be now. And our future plans provide more options.
If you are on windows right now, you are anxiously awaiting the 64 bit V5.5 build, which will get you over the current 32 bit canvas size restriction limitations on windows.
Adding several images. Yes, there are various approaches to this.
You can of course manually swap in different source images as you work. You can use the Recent Source Images menu options to get back to ones you recently used, if you want to swap back and forth.
There are also ways you can load collections of images into a single paint preset. So if you want to create photo mosaic effects for example, you can create an image folder brush, and then paint with the various images inside of the image folder you used for the brush.
You can also work with things like the background texture in the paint synthesizer. That can be an image, or a collection of images, in addition to the procedural options.
I typically work with multiple source images in my Studio Artist artistic work. I use a gallery show feature that lets me setup a collection of images in a folder, and then randomly sequence into them as a part of the gallery show processing cycles.
You could imagine a simple scenario where you have a set of different facial images, and then the painting you create is created from the feature amalgamation of the entire set of facial images (as opposed to just one).
Gallery show can also use the images in a folder like they are frames in a video for processing them with Temporal effects (effects designed to process motion video). This is called stack filtering.
You can also use PASeq Context action steps to hard wire different images specifically to the source area, or to the canvas.
The largest image I had printed was 5 feet tall. The colors were true and it was razor sharp. I had used the super sizer interpolation in Studio Artist to enlarge it. I usually work in SA at 72 DPI or at most 150 DPI and then enlarge it.
Old layers don't need to be gone. I duplicate a layer I consider important and then turn it off. I save my sessions so all the layers are available.
If anyone has the time, can someone 'walk me through' this?
I can only print to about 17x12 inches at home. My printers insist on 300dpi. So, my aim is for at least a 30x20 inch print at 300 dpi. So far I haven't been able to achieve my painterly output (i.e. I haven't gone near vectoring yet) using SA Supersizer alone as it just crashes every time (64 bit running Big Sur on a sizeable Mac).
My process has been, open new source image and canvas. Set canvas to 30 inches on long side, allow SA to calculate short side whilst maintaining proportions, using 72dpi. When my painting is done, I save session, flatten layers then go to Resize canvas, leave dimensions alone but set dpi to 300, crashes every time.
I've had more success by not using resize (its defaulted to Supersizer, by the way) and making any other adjustments in Affinity Photo (I'm non Photoshop) then into On1 Resizer (previously known as Perfect Resize, originally Genuine Fractal) this seems to make a better job than Topaz Gigapixel whilst maintaining the look and quality I want.
Anyone offer help, ideas, am I missing something (apart from the grey cells that disappear with age...)?
I'll run some tests for you later using SA V5.5.
So you want 30x20, at 300 dpi, so 9000x6000, right?
I ran a bunch of 8000x8000 tests a few weeks ago with no crashes doing a wide variety of things. V5.5 of course.
I'll try a few 9000x6000 work scenarios for you and report back on my results.
Bear with me if it takes a few days, i'm kind of overloaded this week.
If you want to suggest a few of the presets you are working with, i can try to address them specifically.
I know at least one of the new video tutorials talks about working with vector effects at smaller work resolutions and then rescaling the output (in Studio Artist) to get to a higher final raster output resolution.
Genuine Fractal or GigaPixel is going to do a better job than SuperSizer at this point in historical time i think. SuperSizer will probably seem overly smooth, which if you a relooking for that would be great, but the other ones are going to be sharper.
We're doing a lot of internal research on neural interpolation, so you might expect a SuperSizer Megatron from us at some point in the future.
I find On1 Photo Raw the best source for enlarging digital images. The resize function may come as an independent plugin.
Thanks Paul. Yes I've found that too; it comes as part of a 'stand alone bundle' as an alternative from the full stand-alone bundle.