OpenTaxa:Guidelines

This page explains how visual assets are made, what types of assets are accepted into the project, and how taxon pages are organized.

The first part of it is structured in the form of a list of frequently asked questions, and further down below, several types of assets are explained with examples.

For potential artist contributors who wish to donate assets but still depend on the sales of their asset collections to generate income, we recommend a system of partial releases similar to the one used in our example Inkwell demo: A commercial release with several types of assets singled out their flora-related images for release under a free license, at once helping out our project and the state of the Commons, and creating a free demo for its project page.

How does the Index work? Why is X species not on it despite having freely licensed photos available?
The Index links readers to species with free license assets available, and when OT was started, projects like Wikimedia Commons and GBIF already featured thousands of relevant files on compatible licenses. However:


 * Manually writing these thousands of species into the Index before starting OpenTaxa's project of asset creation would delay or even hinder OT's primary purpose of stimulating the creation and use of new assets of all types.
 * Thousands of species with only photos would drown out index entries with different types of assets such as 3D scans, game assets and illustrations.
 * A mirror of Commons would just be a worse version of Commons!

Initially, we considered collecting dumps of all pre-existing photos we could find on projects such as Commons, but keeping this mirror updated would be prohibitively time-consuming. After some deliberation, we decided that all individual taxon pages will instead keep organized links to any relevant repositories, and the project can then focus on hosting original asset creations (with a few exceptions, selected from the pool of pre-existing media due to their quality or type): Downloading every single qualifying image from Wikimedia Commons into the OT repository would represent hundreds of thousands of files, while linking to them where they can be easily downloaded, and hosting only the selected most suitable works and our original assets, comes with the best of both options. We also spend a considerable amount of time doing archival work off-site, mirroring free images from smaller websites across databases and repositories like Commons to ensure they are never lost to this project or to the world.

Inclusion of a taxon into the Index must meet at least one of these criteria:


 * At least one type of original asset was created for this taxon by an OpenTaxa contributor
 * 3 or more different types of pre-existing asset were found for this taxon
 * The taxon was requested on the Requests page, and assets for it are an active work in process; therefore a search for pre-existing assets was conducted as part of the creation process

Some assets cannot be hosted here due to file format limitations. However, after a reasonably high amount of original assets are produced, we will soon start to release all assets collected into OpenTaxa on periodical packaged releases (probably on the Internet Archive), as part of our Long Term Preservation project.

What guarantees do you give me about the content you link to?
We are not responsible for external sites, so we cannot make guarantees about the websites we link to. However, collating only free license/copyleft/public domain assets means we can easily only link readers to non-profit websites that are ad-free, friendly towards privacy of their readers, etc. Suitable content on websites with ads will be reuploaded on databases without them before linking. The websites we link the most to include projects by the Wikimedia Foundation, OpenGameArt.org and itch.io, as well as several biodiversity databases including the IUCN Red List, GBIF, and WoRMS. None of the websites previously mentioned display ads. All of them use the HTTPS protocol by default, with the exception of WoRMS, which still uses HTTP, but has a HTTPS implementation. We sometimes also link to websites on the Wayback Machine, which is a part of the Internet Archive. Faithful reproductions of websites on the Internet Archive may include ads.

Notably, original OpenTaxa commissions and creations are currently hosted on users' private accounts, not on an institution account. There is no OpenTaxa account on any of the websites mentioned above. As users create files that are compatible with the project in their free time, such files may be accompanied by other similarly licensed files, or displayed side by side with non-OpenTaxa content in these websites, due to the nature of personal accounts. There is no need to set your account aside exclusively for OT content if you are interested in contributing compatible content.

Feel free to let us know if you spot a broken, dead or hijacked link.

How do individual taxon pages work?
Individual pages listed on the Index should not contain lots of information about the taxon itself, but instead about the assets. All pages start out with a brief lead describing the species' classification, followed by a list of external links where readers can learn more about that species. Read below for reasoning on external links vs. content mirrorring.

After the lead, asset downloads and additional information about the assets can be found, divided in subheaders for each type of asset. This should start with an assessment of the state of a taxon's media availability, written by creators as they work on assets. It may include comments on inaccuracies needing fixes, ideas for variants or improvements, information on resolution, formats etc. This page may also include (preferrably linking to) information on specific aspects of the appearance or behavior of a species as part of guidelines for the creation or improvement of future assets, for example in the form of dichotomous keys to help with telling similar species apart. Links to assets themselves should come after the assessment.

You will notice assessments always mention the month and year where they were written. There is no need to update the "last assessed" date unless the situation for that taxon has changed and a portion of the assessment got rewritten. Otherwise, please leave the date as is for future reference.

Pasting a Wikipedia article on an OT page would always be inferior to linking to the original, as it would need constant work to prevent it from becoming outdated, so we link to relevant encyclopedias, databases and conservation projects such as the IUCN, the World Register of Marine Species, etc. after the assessment of the state of graphical assets and media links. This section can be called "Further reading", to separate it from the content focused on graphical assets.

Finally, pages can have a "See also" section for linking to species that closely interact with it on a given ecosystem. For example, the sword-billed hummingbird has co-evolved with the plant Passiflora mixta, and can only feed on it. If both of these species receive individual pages on OpenTaxa, one should link to the other.

See Limulidae for an example of a page containing all necessary sections, structured in the correct order: Lead, media assessment, media links (Wikimedia Commons category, 3D Scans, OpenGameArt 3D model), then finally further reading (Wikipedia, WoRMS, IUCN, and an identification guide for species)

I created a line drawing from a photo. Can you index it?
As long as the source material is also under a suitable license, or on a license that permits releasing transformative works into a suitable license. These types of drawings are considered derivative works, and we cannot accept derivatives of assets under restrictive licenses. Whenever you create a derivative work, please make sure you send us the source material used as a reference as well. Even if a work's license does not require attribution, OpenTaxa tries to give it wherever possible.

Does OpenTaxa accept fantasy creatures?
We cannot accept fantasy creatures, as OpenTaxa's goal is already very ambitious and must be kept focused on the representation of real species. We recognize the value in creating all types of free assets, including representations of creatures derived from myth or cryptozoology, but we cannot index them.

Are assets based on outdated reconstructions of extinct taxa accepted?
Yes. Even when a reconstruction is proven wrong, it still carries value in depicting the history of paleontology and paleobiology, including the timeline of discoveries that changed how we imagine extinct species today. However, these assets must be clearly labeled as being outdated reconstructions. Examples of outdated reconstructions include dinosaurs as slow-moving lizards, sessile interpretations of Dickinsonia and featherless raptor dinossaurs.

A similar rationale and clear method of error labeling can be used to accept illustrations of extant species that have some innacuracies.

Are cartoony/heavily stilyzed assets accepted?
Cartoony assets can be helpful in school education and related activities, and ideally OpenTaxa would serve all kinds of uses.

The problem is that, many times, these drawings come with simplified or generic features. As an example, let's take a look at the two emblems from U.S. Air Force squadrons below: No amount of speculating could decisively classify the kangaroo further than the Macropodidae family, and in the example of the scorpion, there are literally thousands of matches for its appearance. This is to be expected, as simple, clear designs are a part of stilization. We could add the images to a "Macropodidae (generic)" or "Scorpiones (generic)" pages, but too many generic images and a lack of specific ones is one of the key issues OpenTaxa seeks to address! This isn't to say no generic image should be added - they are useful and they are welcome, but they shouldn't be the focus of OpenTaxa originals, and categorisation of pages shouldn't be done around the expectation of generic body plan images. Images created specifically to portray an entire genera or family of similar species (such as a series of variants on Limulidae) or cartoons created for a specific species (such as this cartoony version of a Xiphias gladius by AntumDeluge) are also good practices. For how similar is too similar to warrant different assets, see the section below. In the example we saw on the beginning of this section, the Kangaroo emblem may be acceptable, as the kangaroo family is relatively small and uniform, but generic drawings for an entire order like Scorpiones may be too generic to be indexed in large numbers - assets should be looked at case by case to see if they are realistically useful or cover new ground.

How specific should assets be?
That depends on how distinctive the species are. In general, scientific illustrations and photography should be classified as specifically as possible, with unidentified species within genera being unadvisable. Assets for non-scientific media such as low-resolution 2d sprites can afford to be made less specific if the species they cover are too similar. For example, no one — not even world-renowned myrmecologists — were capable of telling some species in the Brachymyrmex genus apart without confirmation by molecular analysis. However, we should strive to be as specific as possible, and even minute changes should warrant the creation of asset variants for different species or even subspecies.

In addition to this, a number of factors inside a species are deserving of asset variants. Sexual dimorphism and changes between ant castes are two examples of factors that would warrant the expansion of a species's assets into two or more versions.

An example of assets categorized on a less specific page is Limulidae, as most existing assets depicting horseshoe crabs are generic or uncategorized, and the correction and categorization of these assets is within the scope of the project.

What types of assets can be contributed besides depictions of species?
Assets that are directly connected to a species, but don't depict the species, are acceptable and known as "auxiliary assets". For example, it's easy to see the value in a 3D scan of a beaver's dam on a page about beavers, or a 3D model of an ostrich's egg in a page about these birds. To keep this scope limited, for now OpenTaxa does not index assets depicting human-made objects connected to animals (like drawings of bird feeders or dog collars).

Asset types and contributing
Below we discuss types of assets and their possible applications, as well as types of assets that OT cannot index. Keep in mind OpenTaxa accepts assets even if another asset of the same type already exists for a taxon.

3D high poly models
Models created by 3D sculting, photogrammetry etc. many times have such a large amount of vertices that their use in real-time applications is impossible, but the models can be baked down to the required level of detail with excellent retention of information. For the purposes of OpenTaxa, models remain being considered high poly if they may be too complex for use in real-time applications, even if they have been baked down from an even higher-poly mesh. For example, the Smithsonian's collection of coral scans offers high-poly versions of them with millions of vertices, and low-poly versions that have been baked down to 100.000 faces. Both versions would be considered high-poly for OpenTaxa classification.

Scans
Real objects can be scanned via photogrammetry to generate 3D models that reliably preserve information down to minute levels of detail. The creation of quality scans can sometimes be complicated, requiring expensive equipment and a thorough cleanup process, but the results speak for themselves.

Example: Public domain scans in the Smithsonian Open Access collection.

3D low poly models
For the purposes of OpenTaxa, "low poly" models are all 3D models light enough for use in real-time applications. Either hand-modelled or baked down from high poly meshes, these models are suitable for use in renders, games, etc. Some, after receiving PBR maps, may be indistinguishable at a distance from their high-poly meshes, while others may convey information mostly through flat textures on a very simple mesh (sometimes informally called "PlayStation 1 graphics"). These are also usable outside of stylized graphics, as distant objects on realistic renders for memory and render time optimization. Some 2D assets can be used in 3D spaces with minimal work, such as in the modelling of leaves and relatively flat species. A number of auxiliary assets can be created for these models, including rigs, animations, alternative textures etc. Other types of auxiliary assets are discussed below.

Example: Swamp Limule by CDMir and TinyWorlds.

Photographs
Photos are the most widely available type of public domain taxon depiction. However, even very well-known species can have aspects go completely unrecorded, such as fruit in their unripe stages, or photos showing anatomic details. Still, a staggering amount of species have no publicly available photos, even decades after being described. Good quality photos can also be used to create textures and other types of auxiliary 2D assets described below.

Biological illustrations
2D illustrations created to convey taxon descriptions or behaviors. OpenTaxa divides these between art pieces depicting lifeforms in their natural habitat, and technical illustrations that show taxa with no other elements, in neutral poses and clear proportions, preferably portrayed in some orthographic projection and with indication of scale. Ideally, each taxon would have at least one of each, but currently many described species remain without either type of illustration in the public domain.

2D assets (non-scientific illustrations)
This category includes all drawings created for purposes other than scientific illustration (animation, games, etc.) This includes frame-by-frame animated assets, pixel art, etc. These are usually called "sprites". Note that pixel art and many other forms of 2D drawing need to mind the stilization guidelines. Drawings may be isometric or from orthogonal projections, and can include animations, either as a spritesheet or a pack of single images. Due to the use rationale in this category, a transparent background is recommended, and the use of a lossless compression format is required.

Textures/Materials

 * See also: Physically based rendering on Wikipedia

For the purposes of this project, "textures" or "materials" are all 2D assets meant for use in 3D workflows, as (usually tileable) images that are applied to 3D models. This can include flat textures used to convey color information ("albedo maps") as well as other types of PBR maps, baked down from a high poly mesh or generated from 2D images by some program such as Materialize.