Does AI have a Place in Botanical Art and Illustration?

Written by AI G. Gemini (Artificial Intelligence - Google Genimi)

Edited by a person, Louise Daley

Last month I used AI to assist in making a choice among the many botanical art certificate and diploma courses available. (See Blog HERE.) This month, I decided to ask AI G. Gemini: How can AI be use for traditional botanical art? This was Gemini’s reply: 

In 2026, the intersection of AI and traditional botanical art is not about replacing the artist’s hand, but about augmenting the scientific research, compositional planning, and technical execution of a piece.

Botanical art is uniquely demanding because it requires both aesthetic beauty and rigorous scientific accuracy. Here is how AI is currently being used to support this traditional discipline:

1. Enhanced Scientific Accuracy & Research

Traditional artists often spend hours identifying specimens and verifying anatomical details. AI streamlines this "pre-flight" phase:

• Species Identification: Advanced AI apps like Plantum and iNaturalist now reach over 95% accuracy. They help artists confirm they are looking at the correct subspecies before beginning a week-long painting. 

• Morphology Breakdown: You can ask AI to describe specific botanical characteristics (like "pinnate venation" or "axillary inflorescence") to ensure your sketches are scientifically sound.

• Diagnostic Detail Analysis: AI-driven pathology tools can identify specific diseases or nutrient deficiencies on a leaf, allowing the artist to decide whether to include those "flaws" for scientific realism or to omit them for a "perfect" specimen plate.

• Virtual Dissection: AI models can now generate detailed cross-section references (ovaries, stamens, pollen) from 2D photos, providing a "3D" understanding of the plant's internal structure without always needing to destroy a rare specimen.

2. Reference Image Generation

Traditional artists often struggle to find the "perfect" reference photo. AI can generate custom reference images that fit your specific needs.

• Custom Compositions: If you want to paint a rare orchid from a specific angle with dew drops, you can use models like Gemini or Stable Diffusion to generate a "perfect" reference version of that scene.

• Lighting and Shadow Studies: You can generate an image of a plant with "harsh 45-degree sunlight" to see how shadows fall across complex leaf structures before you commit to a color wash.

3. Compositional Prototyping

Instead of sketching dozens of thumbnails by hand, artists use AI to "test-drive" layouts:

• AI Moodboarding: Tools like Midjourney or DALL-E 3 allow artists to prompt for "18th-century botanical plate composition of a Strelitzia with a landscape background" to see how different balances of negative space and text might look.

• Sketch-to-Render: Platforms like Adobe Firefly or NVIDIA Canvas can turn a rough, 30-second pencil sketch into a semi-realistic color mockup. This helps the artist visualize the final tonal values before committing to expensive watercolor paper.

4. Style and Color Exploration

AI can help you plan your palette and artistic direction without wasting expensive pigments or heavy-grade paper.

• Color Palette Extraction: Upload a photo of your specimen to an AI color tool to generate a "Swatches" list, showing exactly which shades of green and ochre you'll need.

• Style Transfer: Use AI to see how your sketch might look in the style of 18th-century masters like Pierre-Joseph Redouté or Ernst Haeckel. This helps in deciding if you want a minimalist modern look or a vintage scientific aesthetic.

5. Digital-Traditional Hybrids

Many artists use a "Hybrid Workflow" where AI handles the repetitive or technical chores:

• Generative Recolor: Using Adobe Illustrator's Generative Recolor, an artist can scan their line art and instantly see how it would look with different seasonal palettes (e.g., "autumnal decay" vs. "spring bloom”).

• Intelligent Perspective & Symmetry: Software like Clip Studio Paint uses AI to steady "shaky" lines and ensure that complex botanical symmetries—like the fractals in a Romanesco broccoli or the spirals of a sunflower—are mathematically perfect before the artist paints over them manually.

6. Technical Training & Restoration

• Brushstroke Analysis: Some AI-powered brushes in Procreate and Photoshop analyze how a user applies pressure and can mimic the specific "bloom" of watercolor or the "bite" of a graphite pencil, helping students learn traditional techniques through digital simulation.

• Archival Restoration: AI is used to "clean" scans of vintage botanical plates (by Redouté or Merian), removing foxing and yellowing so modern artists can study the original pigment choices and line work with high clarity.

Comparison of the traditional approach and the use of AI

This extensive and detailed list sounds very impressive. However, these examples did not convince me that AI has a place in traditional botanical art. However, rather than explore these programs and making a rebuttal, I ask AI G. Gemini to expose its own flaws. I gave AI G. Gemini the prompt: Make an argument against using AI in traditional botanical art. This was Gemini’s reply: 

The argument against AI in botanical art usually centers on the idea that the practice is as much about the process of observation as it is about the final image. When you automate the "looking," you risk losing the very essence of the discipline.

Here is the case for keeping the "AI" out of botanical art:

1. The Erosion of True Observation

Botanical art is rooted in autoptic evidence—the act of seeing for oneself.

• The "Deep Look": A traditional artist spends hours measuring stems and dissecting blooms. This slow process leads to a profound understanding of the plant’s life cycle.

• The AI Shortcut: AI provides a synthesized "average" of a plant. By relying on a generated image, the artist skips the critical stage of physical interaction, leading to a work that is a "copy of a copy" rather than a record of a living organism.

2. Scientific Accuracy vs. "Hallucinations"

In botanical art, a mistake is not just an aesthetic choice; it’s a scientific error.

• Structural Integrity: AI models often struggle with botanical logic. They might render five stamens on a flower that strictly has four, or attach a leaf to a stem in a way that is biologically impossible.
  

• False Confidence: An artist might trust an AI-generated reference blindly, inadvertently documenting a "hallucination" as scientific fact, which undermines the educational value of the work.

3. The Devaluation of Craft and Soul

Traditional art carries "the mark of the hand"—the tiny imperfections and deliberate choices that reflect a human's emotional response to nature.

• Homogenization: AI tends to produce images that are "perfect" and "smooth," which can lead to a sterile, homogenized style.

• Loss of Intent: Every brushstroke in a traditional piece is a decision. AI-assisted art replaces thousands of micro-decisions with a single prompt, which some argue strips the work of its soul and the artist’s unique "voice."

4. The Ethics of the “Reference"

The debate also extends to where the AI gets its “knowledge."

• Intellectual Property: AI models are trained on the work of thousands of human botanical illustrators without their consent. Using these tools can feel like a betrayal of the community, as the AI essentially "recycles" the labor and style of masters who spent decades perfecting their craft.

I could not have said it better myself.

Argument against using AI by AI

And to clarify the point, (with the assistance of AI) I asked for the most extreme example of AI’s ability by requesting AI G. Gemini to generate a botanical illustration. I entered the prompt: create a hyper-detailed botanical study of Banksia coccinea in the distinct style of Rory McEwen, 20th-century modernism botanical art with high contrast and sharp focus. Leave out descriptions and include the seed pod. I changed the parameters of the prompt several times to get the final image.

Pleasant basic composition, but questionable scientifically inaccurate. This exercise in asking AI G. Gemini to create this composition did make me think deeply about this randomly chosen plant. As I looked at images and illustrations, I realized that only on site observation (Southern West Australian coast) over time will reveal the stages of growth. As I changed the parameters of the prompt, I thought about the best way to compose this plant for artistic rather than scientific purposes, while maintaining scientific integrity. And for posting the results, I imbedded the picture with the words “AI Generated Image” to preserve the integrity of the artist as AI decided to put Roy McEwen’s name to its work.

Banksia coccinea photo from Wikispecies (public domain)

Banksia coccinea by Ferdinand Bauer

Gemini generate image of Banksia coccinea in the style of Rory McEwen