Forest Trail Image by SplitShire from Pixabay

A False Start

Last week, I was discussing an image that I had composited for the pulp campaign with my co-GM.

I made an off-hand remark about how justified I felt about expending the time and effort on the image.

He agreed, positing that the greater the demands on the imagination of a scene, the more useful it was to relieve that load, enabling the players to focus on the game situation.

That sparked a thought that seemed like it would make a good post for Campaign Mastery, so here we are.

My initial reaction – the thought that I mentioned – was that there were two spikes in the utility of graphic depiction – one that focused on the specific but big-picture, and one that focused on isolating a specific from the innumerable possibilities that arose from a more generic subject matter.

The chart to the right describes this first impression, and it’s an insight all on its own that it would be worth exploring in a post.

Unfortunately, a second thought a moment after showed that it’s half-false and half-misleading. There was a critical factor that this alleged relationship was completely ignoring.

A Better Beginning

The missing factor: the potential importance to the plot of a clear representation. This could be said to multiply by the commonality of the subject to give a more reliable index to be charted against the game value of a graphic representation.

So, an image could be of something quite common (a low uniqueness) and even if very central to the plot, it would have a low game value. Or it could be of something quite unique, which means that even if not central to the plot, it has some game value – but if it is central to the plot, its value as an illustration achieves astronomical value.

And, I was all set to write today’s article on this subject – but the presence of one additional factor implied the possibility of more, so before I committed myself, I invested a bit more time in thinking about the subject, and sure enough….

The Final Attribute: Availability

…I found that there was an assumption buried within the analysis to date: that all images desired would be equally available. It took only a moment’s reflection thereafter to realize that this flawed assumption would completely invalidate any analysis that didn’t take it into account, and that there was no one-size fits all answer to accommodate the attribute of availability.

The more I thought about it, the more relevance seemed to attach to this question of availability. At first, I thought there were three types of impact, but further reflection suggested that two of the three were closely related.

Availability Manifestation 1 – The Likelihood Of Success

The first primary impact is on the likelihood that you will find an image matching whatever specifications you might have, or one that is at least close enough.

A secondary outcome would be an image that could be quickly and easily modified to transform it from unsuitable to ‘close enough’ – introducing a sub-variant factor to consideration, which also has to be taken into consideration. The easiest way of doing so is to expand the concept of quality of result to include the potential for modification to achieve suitability.

Availability Manifestation 2a – The Importance Of The Hunt

A problem that needs to be considered is that you can’t know the outcome of a given search until you actually undertake that search. While you can make educated guesses about the likelihood of success based on the specifics of the search, and your capacity to compromise, there are perpetual surprises of both the good and bad kind.

Going on an image hunt is not unlike rolling a skill check – all sorts of things can be reflected in the bottom line, but in the end, it all comes down to luck and the scope for the manifestation of that luck. You can never be completely sure of the outcome until you roll those dice.

‘Availability’ is a critical assessment that can never be perfect. That means that the greater you can refine your estimates of the likelihood of success before you start, and in the early phases of the search, the more accurately you can assess the likelihood of success – and whether or not it’s worth you continuing with the hunt, or should switch to an alternative approach.

Availability Manifestation 2b – The Value Of Image Generation

Not everyone is as adept at image generation as I am. That should surprise no-one. There are also artists out there who can generate works that leave me awestruck in the same time (or less) as it would take me to churn out something just barely adequate.

Also, to be completely honest, sometimes works seem to go a lot more smoothly and quickly than I expected before I started, and at other times, the simplest job seems to have unanticipated complications that add hour after hour to the project.

Again, we have a ‘best guess’ fuzziness attached to each project; you can’t predict in advance with complete precision where that project will fall.

What can be said is that if you honestly evaluate each project relative to an accurate perception of your skills, the actual difficulty and project time will fit more-or-less on a standard distribution, a dumbbell curve, centered on the ‘best educated guess’. In the long run, therefore, your estimates should average out if you are honest with yourself.

But it’s possible to bias the odds in your favor by choosing a base image that is more suited to being manipulated to meet your needs than one that needs a lot more work.

There are all sorts of skills involved, and you get better with all of them with practice – so the bar of what is achievable in a reasonable time-frame keeps rising, and the accuracy of your estimates improves constantly. This means that it doesn’t really matter how adept you are with a (digital) paintbrush; being unskilled simply means that you have more scope for improvement.

Generating an image yourself is always a compromise over finding the perfect image ready-made. Sometimes, that’s a compromise worth making; sometimes, it isn’t – and often, you can’t tell until you are neck-deep in the project.

That’s the point of the (ongoing Image Compositing for RPGs series – collected hints, tips, and tutorials to boost that learning curve into the stratosphere.

“Availability’ is not a simple linear measure of the chance of finding an image that can be modified to your needs; each possible answer also has to be evaluated with respect to the amount of work required, and the amount of prep-time available. It might be that a less-acceptable image with a lower overhead is going to be a better, more practical, choice than one that would yield a better result but would require three times as much work to complete.

Game Value vs Availability

It’s worth taking a moment to define exactly what the goal of the analysis was supposed to be. What I wanted was to develop a schema that identified the relative value attached to searching for, or manufacturing, an image. Such a schema would enable a master listing of priorities that would yield the maximum ‘bang for buck’ for image searches and define a threshold point at which the generation of images (if one could not be found) was worth the time it was likely to take.

I’m not entirely sure that this goal is still a viable one, given the complexities that have now been introduced – but any guideline is better than none. So it’s still worth making the effort. In order to yield a result from these efforts, though, I intend to simplify outrageously. I’m not looking for a definitive set of answers, at least not anymore. A process for making decisions, in which each of the considerations is taken into account, is going to be more universally useful.

Strata of Commonality

If you take another glance at the prototype chart reproduced above, you will notice that Commonality has been divided into four classifications. This still makes a great starting point, so let’s look at each of them.

Mundane

‘Mundane’ incorporates all sorts of everyday items. Most of these will have very limited game utility; what limited functionality they provide can also be achieved through more important images, at least most of the time.

In general, mundane objects should not be graphically represented. It is said that one picture is worth a thousand words – if you can’t see the need for at least 500 words being spent on the description, it’s just not worth searching for them, outside of some specific exceptions.

Vehicles are a special case, so don’t worry about them for the moment. We’re talking pots and pans and treasure chests and the like.

There is a huge contextual element here, however. Space suits may be mundane items in a sci-fi environment, but they would be exceptionally rare in a fantasy one – and, in a steampunk environment, would have a completely different look-and-feel. What’s more, even in a sci-fi campaign, the graphic depiction of a space-suit does so much to ‘sell’ the genre that they have a game value beyond the mere fact of their ubiquity.

Perhaps the greatest value of mundane objects is as ‘set dressing’ to enhance generated images of greater value. These can be transformative – take an image that could belong to any number of settings and toss in a mundane object of greater specificity, and you have suddenly nailed the scene to a far more specific setting.

Throw in an object that clearly doesn’t fit within that range instead, and you introduce a deliberate contradiction that serves as a visual metaphor for a far more complex situation – so much so that the results probably belong in the ‘unique’.range, below.

Common

There are two ways of interpreting ‘Common’. The first deals with generic backgrounds and setting illustrations – these can be useful as foundations of more specific images, or for the capture of genre / setting atmosphere. Or they can impart absolutely nothing of significance. But they tend to be easy to find, so the return on invested time can be relatively high.



Image by ELG21 from Pixabay

For example, a generic image of snow-topped mountains doesn’t add much specific information about a setting – but they carry a sense of grandeur, of sweeping epic scale, that even a thousand words might not be able to convey. Atmosphere alone can justify their use as an illustration.

The other is where the subject is so common that dropping the label is going to result in fifteen different interpretations in the minds of players. For example, “Innkeeper” – everyone will immediately have an image in their reminds as to what he (or she?) looks like, and everyone’s interpretation will be different.

Now, that might not matter – or it might be critical. It all depends on the role of this particular NPC within the adventure.

You can even take advantage of the multiplicity of impressions. For example, you tell the players that they are greeted by an innkeeper in a surly tone, then give each player a strip of paper and ask them to write a line describing what they imagine the innkeeper looking like – race, dress, features.

From that time forward, those descriptions are how each PC sees the NPC – until something happens to reshuffle the strips of paper.

But you don’t tel the players this – you simply incorporate visual elements that no-one else can see into the narrative as the scene proceeds. “He accepts your rebuke of his manners and tips his hat at you in apology.”

“His hat? I thought he wore a leather cap?”

“He does, with a broad rim and silver buckle”….

The players may not know What is going on, but will soon be in no doubt that something more than meets the eye (quite literally!) is transpiring!

Once again, the field can be subdivided, though into more distinct and nuanced classifications than just “generic” and “atmospheric”. I’ll have more to say about that when I get into specifics a little later in the article.

Specific

The more specific your search subject, the lower the likelihood of an exact match, and the more you either have to compromise or edit whatever you find to make it fit for purpose. In fact, the likelihood of any result is markedly lower.

That can be compensated for, to some extent, by making more exhaustive searches. You can also sometimes find results that won’t show up in Google Search results by trying a different search engine – I’ve noticed that they usually all give different overall results with a few images in common. My search priority these days is usually DuckDuckGo, Google, and Bing (in that order) and I’ve recently added a fourth string to my bow, Qwant.

After that, some of the big image repositories and clip art providers, like Pixabay sometimes have images that none of the search engines seem to find. I keep a large set of links to these for the purposes of searching out illustrations for Campaign Mastery, anyway. You might not have any equivalent justification, but don’t let that stop you. Avoid (as much as possible) sites that watermark their free images, though. I include Wikimedia in this category.

The deeper that your search has to proceed, though, the more you encroach upon the point of simply not finding what you are looking for. Being aware of this, I also open any images that could potentially be edited to give the desired result, preparing the ground for a potentially-necessary Plan B.

It is also important, when making such assessments, that the points made earlier are kept in mind – “is it actually worth the investment in time to edit this image into something with game value?” – the more time you spend in the search, the less time there is available for such editing; there comes a point where the answer to the italicized question becomes “no.”

Unique

This is really an extension and extrapolation of the trends identified in the last category; the only real change is that the Game Value increases massively by virtue of the “unique” label.

When it comes down to it, with very very few exceptions, “Unique” images follow one of two paths:

• You perform all searches with the expectation that whatever you find will require editing to convert potential game value into actual game value; or
• You choose the best image that you can find and modify the adventure to fit, not the other way around.

In terms of time, choice two is the obvious winner. So if I find an image that will “work” with this approach, that’s job done and move on to the next; if I don’t, then I’m looking at the ‘editable’ options that I have gathered in the course of the search. Pursue both options simultaniously, in other words.

Classification of Image Categories

Okay, let’s get down to brass tacks. I’ve divided image subjects into six categories. I’m then going to look at how each category intersects with the Strata of Commonality just discussed, permitting advice that is as specific as I can make it. The six categories are:

1. Objects
2. People (NPCs)
3. Monsters & Encounters
4. Vehicles
5. Locations
6. Events & Effects

There is some overlap – “Unique People” clearly overlaps with “Encounters”, for example, and “Objects” can include “Vehicles”. These overlaps were necessary to ensure comprehensive coverage.

Objects Vs Vehicles

If the PCs involvement / engagement with the vehicle is enough that it may require depiction of the interior, then the “Vehicle” category is the important one, with the “Object” treatment (exteriors only) a fallback position.

If there is not likely to be any Game Value in the interiors, then the Vehicle is an “Object”, and the advice attached to that category should be your guide.

Vehicles like motorcycles that don’t have an “interior” require a further exercise of mental gymnastics – pretend that the vehicle in question is actually one that has an interior and then assess the engagement as above.

People Vs Encounters

The “People” category has two non-exclusive objectives – implying the personality of the individual, and providing a common mental image to the players to aid in recognition of the NPC as an individual. If neither of those is the purpose of the image, it should be treated as an “Encounter”.

Encounters are more related to what the individual is doing in the image, which should match up with the plot purpose of the encounter. “Why is this encounter happening?” at a meta-plot level is the guiding principle.

I want to conclude this overview section with a comment about player agency. Giving players a choice is always a better choice than not, but it does make image searches more complicated. For example, in the Zenith-3 campaign, the plot had the players buying a couple of vehicles from amongst the options at a pair of used-car lots. What attributes the players chose to prioritize would dictate which options best matched; since I didn’t know this, I had to prepare for a large number of contingencies. I described the situation and the prep involved in How Good Is That Rust-bucket In The Showroom Window?

There were more than 200 vehicles in the spreadsheet that contained the results. After the players decided on their key parameters, from highest to lowest, I simply had to sort the results accordingly to determine which vehicles best matched. Then it was just a matter of playing around with the budgets to derive a set of choices. I could simplify this somewhat because these were to be bought from used-car lots that were no doubt trading in vehicles even as the PCs were searching – any option that yielded too many possible combinations or that didn’t look to be “fun” / interesting, I simply marked as “sold” while the PCs were looking around the lot.

Because I could perform “theoretical” sorts in advance, I could manifest a shortlist of the options that would represent the options to be put before the players. This ended up being a list of twelve vehicles; there were more, but another key metagame factor was whether or not I could find an appropriate or editable image of the vehicle.

For those who may be interested, the twelve second-hand cars offered to the PCs in post-Ragnarok 1986 were:

• 1983 Black Coupe DeVille Cabriolet d’Elegance
• 1982 Sky-Blue and Burgundy Sedan DeVille
• 1984 Black Escort LX 5-door Hatchback
• 1984 Beige Escort Series I Liftback Wagon
• 1984 Black Escort 3-door Hatchback
• 1984 Navy Blue Cadillac Sedan DeVille Automatic
• 1983 Lincoln Continental Mark VI Pucci Designer Edition – one of the two chosen
• 1984 Brown Chevrolet Cavalier Station Wagon
• 1983 Cherry Red Chevrolet Cavalier, badly faded
• 1982 Purple Buick Skylark Station Wagon – the second vehicle chosen, but it turned out to be a good-looking lemon
• 1983 Bright Red Chevrolet Cavalier – the vehicle chosen as a replacement for the Lemon above.

Each of these had their virtues from the PCs perspective. That meant that each had to be presented to the players when they were looking at their different purchase options – I had no idea which ones the players would eventually choose. To get two cars, I had to illustrate 12 (plus three interior views).

That’s a lot more work than shows in the ultimate results. Player Agency is the enemy of efficient prep – but that’s a necessary evil. The only real restrictions placed on the players choices were (1) that I could find / make a suitable illustration of the vehicle; and (2) that I wanted them to end up with two different makes and models so that I could compare and contrast the two, and so that the players choices had measurable impact on the game-play. Because if it makes no difference, it’s not really player agency, is it?

Where To From Here?

This article is now approaching an unmanageable length, given the available time. So I’ve decided to break the rest of it into smaller mini-posts, which I’ll deliver over the next week or so, a day or two apart. Each will examine one of the specific Image Categories listed above, breaking each down into the four levels of Commonality.

Image by Gerd Altmann from Pixabay