I've been dinking around with S.A. for a number of years, and have just now gotten more serious, and I'm posting photographs to an online gallery. I have some fundamental questions to which I've yet to find answers. Since my knowledge base is shallow, please bear with me. This might get difficult.
1. What can I do to tweak my photos so that I get quality that will be sustained if my photos are enlarged to around 20 inches in width/length? By changing the canvas size and DPI, am I making the photos good for larger prints? I use a Canon SX60 SuperZoom for most of the photos, and I sometimes use an iPhone 12.
The standard DPI for SA is at 72. If I increase that figure to 300 DPI, or even 900 DPI, does that adaptation translate into sharp quality photos if enlarged? I don't want resolution loss if someone is interested in purchasing a photo and the resolution decreases to the extent that photo quality suffers.
Once I start processing photos with SA, am I leaving the photo data that comes from my camera behind and am I now working only with SA parameters? Does that mean that the SA files are independent of the camera data?
BTW, I am enjoying the bejesus out of SA, even though I know that I know virtually nothing about the areas I haven't used. And that's a lot. So I need some fundamental information about the relationship between the SA data and camera data. Can anyone help?
Thanks so much in advance.
Replies
So i guess i should ask if there are camera meta data tags that we don't pass along that maybe we should have an option to? I honestly don't know or have any personal interest in it. But if people feel there is something important there we are currently missing feel free to speak up.
The other part of that is whether we should be tagging our output raster image files with some kind of color-metric data tags?
i always use 300 dpi if i am selling and/or printing out a print because that's what i always read i should do. then i get confused when the size changes. there are places online using a.i. to resize photos but i've gotten good results with the supersizer in s.a.
i recently supersized an old home movie that was on a vhs tape. it was only one minute so i resized it with paint and vector presets. i will not be doing that for the 8 minute ones. a few weeks ago i was horrified to find that would i recorded on zoom was output at 640X360 because it was long. i supersized it with a paint sketch effect and i will probably use it.
Oh, jeez. I know so little. I don't know what the supersizer is...
more study...
Thanks.
i think i made a screenshot of the supersizer.
The only resolution factor that matters in a raster image file is the number of pixels.
You can think of the dpi tag attached to the file as metadata that maps it to some fixed display dimension in a print. But you can change that dpi metadata anyway you want, and the size of the print associated with it will change accordingly. The number of pixels isn't changing when you do this, just the size of the display mapping.
Studio Artist is no different than Photoshop in this regard. In both programs, you can go into a resize dialog and just change the dpi tag without changing the number of pixels, or you can change the number of pixels in the image.
When you change the dpi tag associated with a fixed size pixel image, the 'size' of the displayed image in cm or inches changes to match the change made in the dpi tag. But the number of pixels in the image remains the same.
If you change the actual number of pixels, then the display size of the image at a given dpi tag setting will change. Increasing if you increase the number of pixels, decreasing if you decrease the number of pixels.
The StudioArtist canvas is just a raster image buffer. There may or may not be an associated set of bezier paths tagging along with it in the bezier path frame tied to a Studio Artist layer. But when you save the canvas as a raster image, you are outputting a raster image file. You can associate any dpi tag you want to that pixel data. But the pixel data is the important part, you can arbitrarily change the dpi tag in Studio Artist or in Photoshop at any point afterwards without changing the pixel data.
Some Studio Artist effects are vector in nature, and can be output as svg vector files. The svg vector file is resolution independent. So if you are working with vector effects in Studio Artist, you could work in a 1024 pixel canvas to get your design finished, then output it as a svg vector file, then import that back into Studio Artist into a 8000 pixel canvas if you wanted that kind of pixel resolution for your final print, then output that 8000 canvas out as a 8000 pixel raster image file.
Again, the dpi tag associated with that 8000 pixel raster image is just a meta data tag. You can arbitrarily change it in either Studio Artist or PhotoShop to be whatever you want it to be (without changing the pixel data at all). The associated display size for the dpi you specify will go up or down appropriately as you do that.
You can import svg vector files directly into Illustrator and PhotoShop as well, rendering them into whatever raster pixel size canvas you desire.
Another approach to working in StudioArtist is to build your effect composition at a lower resolution, then rerender the canvas and the associated PASeq or History Sequence to the final higher resolution you want. Then you playback the rerendered PASeq or History Sequence and all of the effects are run again into the higher resolution canvas.
How well rerendering the PASeq or History Sequence works is a function of what effects you are using. It might work great, or it might have issues. You really need to experiment with it for whatever visual effect approach you are using to see if it makes sense for your particular workflow and artistic effects or not.
We've been improving the rerender process for the PASeq and History Sequences over time. And we have a number of ideas to improve it more in the future.
Some StudioArtist effects are essentially resolution independent. Many more are not. And typically people mix and match them within the same workflow for a given artistic project The re-render process tries to adjust the parameters associated with them if they are not resolution independent so they scale in an appropriate way when the canvas is resized and the recorded effects (as action steps in the PASeq) are rerendered.
Often the rerender process is kind of like an edition print. So the overall effect might be similar, but if you look closely individual details will be different. Some of that is due to the nature of how the random noise generators used through different StudioArtist effects work. But even if you setup the random number seeds to be the same, as you increase the size of the canvas, different interactions going in a particular effect (or how it interacts with other things in the canvas) might change. It's like the butterfly effect in chaos theory if you are familiar with that.
It used to be that using interpolation to resize the number of pixels in a raster image lead to very noticeable interpolation artifacts. Different interpolation algorithms have different characteristic artifacts. So artists felt like they had to work at their final desired pixel size for their canvases to avoid the issues associated with interpolation artifacts lowering the quality of their final work.
However, the state of the art in this area is rapidly changing for the better with recent developments. Things like neural net interpolation (and fractal interpolation) can dramatically improve the results of pixel rescaling interpolation algorithms. So your thinking on this kind of approach might need to be rethought if you are basing it on how things worked and looked visually when bilinear splines were your only option.
My personal viewpoint on this is that things are very quickly moving to the point where one would want to work with lower resolution working canvases to build up an effect, then use a super high quality interpolation algorithm to resize that to your final print requirement.
And if your work is all vector based, then again, you want to be working at a comfortable lower res working resolution to put your effect together, then rerender that vector information at the final high resolution you want for you final output after the design work is finished.
One of the most common mistakes people make in Studio Artist is working with a fully vector effects like the Vectorizer, and assuming they need to feed it super high res images, and render it in a super high res canvas. In general, that is absolutely the wrong approach to take. It will dramatically slow down render times, and will lead to worse looking results as well. You are much better off for many different reasons working in a much smaller res canvas (like 1920 lets say), then outputting svg and then rerendering that to your final desired res for the final output.
Our plan in V6 is to incorporate something built into the Vectorizer that essentially does that for you even if you insist on running the vectorizer into super high res canvas sizes. That future enhancement will internally work at lower res so that the existing 'generate svg, output svg, rerender svg into a higher res canvas' workflow will be done for you automatically by the program behind the scenes. So that is the plan. But today in V5.5 you need to be aware of this.
The V6 plan is to also build much more sophisticated interpolation algorithm options directly into the program for raster canvas resizing. If you are working in V5.5 today, and want to use a more sophisticated raster interpolation algorithm to resize your raster work than what we offer, there are various 3rd party products out there that give very good results.
Studio Artist V5.5 is 64 bit on mac and windows. So you can work directly in high res canvas sizes if you absolutely need to. And sometimes that might be the correct approach to getting a certain thing done. But oftentimes it isn't, as i detailed above.
In general, experimentation is always a good approach for different effect scenarios. Try out some test experiments, see what works for you, what doesn't.
If you run across rerender issues you think should play out better, send us the PASeq with a description of what you were trying to do, and we might be able to make individual action steps in the PASeq rerender better in the future.
Companies like apple will always tell you you need more and more resolution to do 'pro' work, because they want to continuously sell you new machines and higher res displays every year.
We're trying to rethink how people work. And hopefully come up with workflows that enable you to do your design work at lower res, and then have good approaches to scaling that to whatever final output res you desire. We still have some work to do to get that to the place we envision it is going to be at, but i think it's definitely going to come together fully in the next year or so.
Thanks for the time you took to write this detailed reply. I'll set aside some time to try to absorb all of it.