This series is fascinating. I'm using a folder of black ink splotches (photo of ink thrown at a white page). So the source photos only have black or white in them. I'm also using one of them as the loaded style image. My Studio Artist V5.5 generative strategy is pretty elaborate (will explain more in a post). I love how it starts creating color and structure that extrapolate off of those inputs.
Read more…
Ok, let's dive into what is going on in the generative strategy for this particular set of imagery.
So I'm using images that were created from black ink dripped or thrown onto a white paper. Like the image below.
I put a bunch of them into a folder. I'm then using that as my source folder in gallery show.
I then concocted a generative paint strategy in gallery show.
So I'm mutating the factory paint presets (mutating across the entire range of paint presets in my preset folder).
The 2 options below (Paint Transform and Paint Draw Option) are generative options I can turn on. I selected the make MSG BrushLoad option for Paint Transform. This runs a Smart Edit command that auto-edits every mutated paint preset gallery show generates, editing it so that it is using the MSG Brush Load features. A random MSG preset from the entire MSG factory set is selected to be the MSG running inside of Msg Brush Load.
I selected the End IpOp PathStart Sketch option for the paint Draw Option. What this does is randomly generate an IpOp effect that generates bezier path output. Those bezier paths are auto-loaded into the current paint preset (already heavily auto-edited by gallery show). The Ip Op effect is embedded into the paint preset (so the bezier paths generated by the randomly generated Ip Op effect processing the current source image are used as the paint paths for the paint preset).
Studio Artist V5.5 also has generative paint preference options, located in the new Generate tab in the preferences dialog. These also affect any paint presets coming out of a gallery show technique.
As you can see, there is a lot going on in here with this particular set.
The Begin, Mid, Mid2, and End Paint Options are all affecting any paint presets coming out of a gallery show technique. You can select different Smart Edits that will be run at the appropriate place in the generative contraction sequence begin run in gallery show for paint presets.
The sequence of generative edits is
Technique - Begin - Paint Transform - Mid - Mid2 - Paint Draw Option - End
As you can see, you can really dive in and customize all of these different options. Or you can set them all to None, and just live with the output of the gallery show technique without any additional generative smart edit options.
So let's take a look at some of the output I generated during this little experiment last night while I was testing the program.
So this first one shows off one of the randomly selected ink splotches repainted using an automatically generated MSG Brush Load paint preset.
Here's the next one in the sequence (I'm running gallery show continuously in cycles, grabbing the output by manually hitting the Grab button in the gallery show toolbar to save the canvas when I see something I like).
I'm never erasing the canvas, always overdrawing. However, I am modifying the canvas each gallery show cycle (more on that later). I'm also intelligently auto-masking (more on that later as well).
Note how this second new preset derived from mutating existing paint preset followed by a set of generative smart edits that convert it to use MSG Brush Load may start to introduce color. If you look up at the Palette Option in the Gallery Show Preference tab screen shot above, you can see I have it setup to randomly generate a new color palette and set of source area color memory for each gallery show cycle.
Let's move to a little hit later in the gallery show run.
You can see that it's starting to get a little bit messier. You can also see that there is this interesting thing going on where the pre-existing canvas painting seems to be transforming as well. So this generative strategy I'm using also modifies the last canvas output each GS cycle.
The Rand option for the AutoMask Src (gallery show preferences) randomly selects either the source image or the canvas as input to the intelligent automatic selection mask being generated each gallery show cycle. There are lots of different Auto Mask Options, I'm using the Rank Edge one for this little experiment. The Rank Edge IpOp effect is randomly mutated in heuristically intelligent way to derive a selection mask off of whatever input was randomly selected for it. The output of the Ip Op effect is routed to the Selection buffer. The gallery show technique is then auto-masked by whatever is in that selection buffer. Since it's paint, it could be using any of the different paint section mask options (again randomly selected for you).
Let's continue on for a few more gallery show cycles we examine this weird little generative paint experiment I ran last night.
I choose these particular examples to help give you some sense of how the generative paint strategy is both piling on additional patin to the canvas while at the same time transforming the actual complete canvas image itself each gallery show cycle.
Let's take a look at the start and end cycle processing being used in gallery show for this experiment.
So my start cycle processing is using a randomly selected preset from the current Favorites preset category. I set that category up to contain a bunch of different Vectorizer presets that process the canvas (not the source). If they were not setup to all process the canvas (probably true for all of your vectorizer preset), you can change the popup that says Default on the right side of the screen shot above to the Force Canvas option. The preset being used for the start cycle processing will then be auto-edited on the fly (the instance we are using in GS, not the actual preset on disk) to process the canvas instead of the source.
For my end cycle processing, I'm using the generative Random AdaptGradLighting option. This automatically generates a randomized gradient lighting preset in conjunction with a randomized adaptive contrast adjustment preset for the end cycle processing (essentially it auto-generated a PASeq for you that does the end cycle processing). A new unique one will be auto-generated each GS cycle.
So I realize that was a ton of new information to take in. Probably a little bit over-whelming at first.
We used gallery show to create a generative strategy that automatically generates a new unique paint preset that embeds both another new unique MSG preset and another new unique Ip Op preset into the newly created paint preset.
We do this every gallery show processing cycle.
We also randomly generate a new source color palette and set of source color memories each GS cycle.
Esch GS cycle randomly selects a new source image from a folder of images you curated, automatically generates the Brain teaser paint preset that uses the paint synthesizer, Msg, and Ip Op features in one paint preset.
A randomly selected Vectorizer preset from a favorites category is used for GS start cycle processing to vectorizer the last canvas output image.
A selection mask is auto-generated each GS cycle, and used to mask the GS Technique (the paint preset it generated in this case that is sued for the painting into the masked canvas).
A randomly generated PASeq is then used for GS end cycle processing. It adaptively adjusts the canvas contrast as well as adds gradient lighting to the canvas.
All of that is one gallery show cycle.
We then do it over and over again.
I'm sure this particular example is too abstract for some people, so we'll do this again for something more representational in another example.
After you get exposed to this a few times, you will see that it's not really that complicated.
You choose a gallery show technique.
You select any additional options that might be available for the technique you selected.
You choose a source folder of images you want to work with.
You choose any auto-masking options you want.
You choose any start cycle and end cycle processing you want.
You let it run, or you manually run it in single cycles.
Comments
Ok, let's dive into what is going on in the generative strategy for this particular set of imagery.
So I'm using images that were created from black ink dripped or thrown onto a white paper. Like the image below.
I put a bunch of them into a folder. I'm then using that as my source folder in gallery show.
I then concocted a generative paint strategy in gallery show.
So I'm mutating the factory paint presets (mutating across the entire range of paint presets in my preset folder).
The 2 options below (Paint Transform and Paint Draw Option) are generative options I can turn on. I selected the make MSG BrushLoad option for Paint Transform. This runs a Smart Edit command that auto-edits every mutated paint preset gallery show generates, editing it so that it is using the MSG Brush Load features. A random MSG preset from the entire MSG factory set is selected to be the MSG running inside of Msg Brush Load.
I selected the End IpOp PathStart Sketch option for the paint Draw Option. What this does is randomly generate an IpOp effect that generates bezier path output. Those bezier paths are auto-loaded into the current paint preset (already heavily auto-edited by gallery show). The Ip Op effect is embedded into the paint preset (so the bezier paths generated by the randomly generated Ip Op effect processing the current source image are used as the paint paths for the paint preset).
Studio Artist V5.5 also has generative paint preference options, located in the new Generate tab in the preferences dialog. These also affect any paint presets coming out of a gallery show technique.
As you can see, there is a lot going on in here with this particular set.
The Begin, Mid, Mid2, and End Paint Options are all affecting any paint presets coming out of a gallery show technique. You can select different Smart Edits that will be run at the appropriate place in the generative contraction sequence begin run in gallery show for paint presets.
The sequence of generative edits is
Technique - Begin - Paint Transform - Mid - Mid2 - Paint Draw Option - End
As you can see, you can really dive in and customize all of these different options. Or you can set them all to None, and just live with the output of the gallery show technique without any additional generative smart edit options.
So let's take a look at some of the output I generated during this little experiment last night while I was testing the program.
So this first one shows off one of the randomly selected ink splotches repainted using an automatically generated MSG Brush Load paint preset.
Here's the next one in the sequence (I'm running gallery show continuously in cycles, grabbing the output by manually hitting the Grab button in the gallery show toolbar to save the canvas when I see something I like).
I'm never erasing the canvas, always overdrawing. However, I am modifying the canvas each gallery show cycle (more on that later). I'm also intelligently auto-masking (more on that later as well).
Note how this second new preset derived from mutating existing paint preset followed by a set of generative smart edits that convert it to use MSG Brush Load may start to introduce color. If you look up at the Palette Option in the Gallery Show Preference tab screen shot above, you can see I have it setup to randomly generate a new color palette and set of source area color memory for each gallery show cycle.
Let's move to a little hit later in the gallery show run.
You can see that it's starting to get a little bit messier. You can also see that there is this interesting thing going on where the pre-existing canvas painting seems to be transforming as well. So this generative strategy I'm using also modifies the last canvas output each GS cycle.
The Rand option for the AutoMask Src (gallery show preferences) randomly selects either the source image or the canvas as input to the intelligent automatic selection mask being generated each gallery show cycle. There are lots of different Auto Mask Options, I'm using the Rank Edge one for this little experiment. The Rank Edge IpOp effect is randomly mutated in heuristically intelligent way to derive a selection mask off of whatever input was randomly selected for it. The output of the Ip Op effect is routed to the Selection buffer. The gallery show technique is then auto-masked by whatever is in that selection buffer. Since it's paint, it could be using any of the different paint section mask options (again randomly selected for you).
Let's continue on for a few more gallery show cycles we examine this weird little generative paint experiment I ran last night.
I choose these particular examples to help give you some sense of how the generative paint strategy is both piling on additional patin to the canvas while at the same time transforming the actual complete canvas image itself each gallery show cycle.
Let's take a look at the start and end cycle processing being used in gallery show for this experiment.
So my start cycle processing is using a randomly selected preset from the current Favorites preset category. I set that category up to contain a bunch of different Vectorizer presets that process the canvas (not the source). If they were not setup to all process the canvas (probably true for all of your vectorizer preset), you can change the popup that says Default on the right side of the screen shot above to the Force Canvas option. The preset being used for the start cycle processing will then be auto-edited on the fly (the instance we are using in GS, not the actual preset on disk) to process the canvas instead of the source.
For my end cycle processing, I'm using the generative Random AdaptGradLighting option. This automatically generates a randomized gradient lighting preset in conjunction with a randomized adaptive contrast adjustment preset for the end cycle processing (essentially it auto-generated a PASeq for you that does the end cycle processing). A new unique one will be auto-generated each GS cycle.
So I realize that was a ton of new information to take in. Probably a little bit over-whelming at first.
We used gallery show to create a generative strategy that automatically generates a new unique paint preset that embeds both another new unique MSG preset and another new unique Ip Op preset into the newly created paint preset.
We do this every gallery show processing cycle.
We also randomly generate a new source color palette and set of source color memories each GS cycle.
Esch GS cycle randomly selects a new source image from a folder of images you curated, automatically generates the Brain teaser paint preset that uses the paint synthesizer, Msg, and Ip Op features in one paint preset.
A randomly selected Vectorizer preset from a favorites category is used for GS start cycle processing to vectorizer the last canvas output image.
A selection mask is auto-generated each GS cycle, and used to mask the GS Technique (the paint preset it generated in this case that is sued for the painting into the masked canvas).
A randomly generated PASeq is then used for GS end cycle processing. It adaptively adjusts the canvas contrast as well as adds gradient lighting to the canvas.
All of that is one gallery show cycle.
We then do it over and over again.
I'm sure this particular example is too abstract for some people, so we'll do this again for something more representational in another example.
After you get exposed to this a few times, you will see that it's not really that complicated.
You choose a gallery show technique.
You select any additional options that might be available for the technique you selected.
You choose a source folder of images you want to work with.
You choose any auto-masking options you want.
You choose any start cycle and end cycle processing you want.
You let it run, or you manually run it in single cycles.
Very informative post and cool images, thank you!