Research & production log · updated Oct 21, 2026 · 00:48

Bad Buddhist Radio Lab

Turning the 2006 Bad Buddhist Radio podcast into chapters, blog posts, Remotion videos and airbrush illustrations. Episodes 1–3 come first; the other 54 wait for your go-ahead. This page also records which airbrush models are worth using.

Credits used 626 of 4808Balance 4182Images generated 197Episodes gated at 3

StatusWhere each episode is

EpisodeTranscriptChaptersBlog postImagesVideos
01 Just Some Dude Who Woke UpDoneDoneDoneDone6 videos
02 Get Out of Your BubbleDoneDoneDoneDone6 videos
03 All Life Includes SufferingDoneDoneDoneDone6 videos
04–86 54 more episodesWaiting on youWaiting on youWaiting on youWaiting on youWaiting on you

FindingsWhat the research says so far

  1. Most create "engines" are one backend with a style hint.

    All 15 engines rendered text cleanly, including the SD 1.5 names (Anything V6, Protogen, Wifu), which real SD 1.5 models cannot do. DALL-E 3, Slow Picasso, SDXL Pro and Flux produced the same composition for the same prompt (layout correlation 0.80–0.82). The only visible differences are style flavors: Protogen adds sci-fi, the anime names add anime, 3D Diffusion adds clay.

  2. Seeds do nothing, and unknown engine names still work.

    The same prompt, engine and seed gave unrelated images twice (correlation 0.03). A made-up engine name (zz_fake_engine) returned an image from a fallback model. So consistency across a series has to come from reference images, not seeds.

  3. Price, not quality, separates the create engines.

    On a 1344×768 image, fl_flux_pro2, fl_flux, gpt_image_15 and gpt_image_2 cost 1 credit. Midjourney and every legacy n_* engine cost 2, and DALL-E 3 costs 19, with no visible gain. Use the 1-credit engines.

  4. Diagrams and rough layouts beat text-only prompts.

    Feeding a hand-built SVG diagram or blocking sketch to an edit model preserved the composition exactly and every label word for word, including the small captions. Text-only prompts got the four sights roughly right. Sketches got the figures in the exact order and place.

  5. Only two edit models return 16:9.

    Nano Banana (gemini_flash) and GPT Image 2.5 return 2016×1152 from a 16:9 input. Flux.2, Kontext, GPT Image 1.5 and Seedream return a 1600×1600 square. For video frames, use the 16:9 pair.

  6. GPT Image 2.5 can create from scratch through a blank canvas.

    create --engine gpt_image_25 falls back to a generic model. An edit on a blank 16:9 canvas runs the real thing: best prompt adherence of any run, clean titles, 2048×1152, 5 credits, about 35 s.

  7. The prompt matters more than the engine.

    The hero formula from the research guide (subject on the right third, empty space for a headline, a 3-color palette, "No text") worked on every engine and produced usable blog headers on the first try.

  8. Gallery search is a stock-photo search.

    airbrush search returns Pexels photos, not AI art. Short noun queries ("koi pond", "ornate indian palace gate") work well. Scene descriptions don't ("village lane at dusk with oil lamps" returned only lamps). It's useful for real reference photos.

  9. One style-anchor image holds a whole series together.

    Passing the gouache four-sights image as "style reference only" produced new scenes (the ascetics, the lane cutter, the row of doors) that look like one illustrator's set. Combining a layout sketch with a photo of a real Jaipur gate in one edit gave both exact blocking and real architecture.

  10. ElevenLabs v4 is live and handles the expressive tags.

    The account lists eleven_v4 (10,000 characters per request). Test clips with [mischievously], [sarcastic] and [laughs] came back without any tag words in the Spokenly transcript, so the tags are performed, not read. Nine narrations (3 voices × 3 episodes) came out between 116 and 127 seconds.

PipelineWhat to use for each job

JobMethodCreditsWhy
Blog hero imageHero-formula prompt → gpt_image_25 edit on a blank 16:9 canvas5Best adherence, true 16:9 at 2048 px, leaves room for the headline.
Cheap drafts and variationsHero-formula prompt → create with fl_flux_pro2 or gpt_image_2, wide_landscape1Same backend quality as the 2-credit engines at half the price.
Concept diagram (truths, wheel, path)Hand-built SVG → gemini_flash or gpt_image_25 edit, labels spelled out in the prompt5Keeps the layout and every label; 16:9 output.
Story scene with a specific compositionRough blocking sketch (16:9) → gemini_flash or gpt_image_25 edit5Exact control over who stands where.
Consistent seriesOne style-anchor image passed as Image 1 in every edit5Seeds are ignored, so a reference image is the only way to hold a style.
Square social cardsSame diagram/sketch → flux_pro2 edit4Returns 1600×1600 squares.
Avoiddalle3, seedream, n_protogen-4.5, n_anything-v6, n_wifu-diffusion–Cost more for nothing (19 and 7 credits), or force an unwanted style.

EnginesIs each model worth it?

Credits are measured from balance changes on 1344×768 creates and default-size edits. Speed is the mean wall time per image.

EngineCreditsSpeedWhat we sawVerdict
fl_flux_pro2
Flux.2 Pro · create
111 sDistinct painterly look on 2 of 6 prompts, clean infographic, legible title.Use
gpt_image_2
"Google Text Master" · create
113 sDramatic compositions; bold saffron ink style on the hero prompt. Title partly hidden behind the figure once.Use
gpt_image_15
"OpenAI Davinci" · create
114 sCleanest title text. Infographics came out small with too much empty space.Use
fl_flux
Flux · create
114 sIndistinguishable from the shared pool; good cheap option.Use
n_midjourney
Midjourney V7 · create
216 sGood atmosphere; best four-pane pond metaphor. No sign of real MJ behaviour, at twice the price.Situational
n_sdxl · n_stablediffusion · n_slow_picasso · n_analog · n_3d · n_realistic-vision
legacy names · create
25–18 sSame backend with mild style hints (photo, clay, sketch). Nothing the 1-credit engines can't do with one adjective.Situational
n_protogen-4.5
Protogen · create
29 sForces sci-fi (a spaceship door on the Buddha's palace). Garbled title.Skip
n_anything-v6 · n_wifu-diffusion
anime names · create
24–6 sForces anime and adds unasked-for captions.Skip
dalle3
DALL-E 3 · create
1918 sSame composition as the pool at 19× the price.Skip
gpt_image_25
GPT Image 2.5 · edit
523–35 sBest adherence; 16:9 output at 2016–2048 px. Also creates from scratch via a blank canvas.Use
gemini_flash
Nano Banana · edit
522 sKeeps 16:9 (2016×1152), exact labels, fast.Use
flux_pro2 · flux_kontext · gpt_image_15
edit
420–30 sExcellent layout and text fidelity, but output is a 1600² square.Situational
seedream
Seedream · edit
720 sSame results as the 4-credit edits; square output; most expensive edit.Skip

ExperimentsRounds run

Round 1 · 182 credits · 86 images

Engine screen: one prompt, every engine

Question
With identical prompts, which of the 15 create engines gives the best images for these episodes?
Setup
Six prompts (a painterly scene, an abstract metaphor, a cartoon, a photo, a title card with text, an infographic) × 14 engines at 1344×768, plus DALL-E 3 on two.
Verdict
The engines are far more alike than their names suggest. Choose on price and pick the 1-credit ones.
  • Text rendered on all 15 engines; the usual title-card failure was the figure's head covering letters (8 of 15).
  • Output size and format switch between 1344×768 WebP and 1920×1280 JPEG regardless of engine, which suggests two providers behind one pool.
  • Protogen and the anime names inject their style even when it fights the subject.
c1 · The Four Sights (gouache). Many engines return the same composition: prince in a carved window at left, procession in the middle.
c1 · The Four Sights (gouache). Many engines return the same composition: prince in a carved window at left, procession in the middle.
c2 · The Four Ponds. Seven engines independently drew the same captioned 2×2 grid.
c2 · The Four Ponds. Seven engines independently drew the same captioned 2×2 grid.
c3 · Bubble traffic (cartoon). The most varied prompt; Flux.2 Pro and SD Pro added funny speech bubbles.
c3 · Bubble traffic (cartoon). The most varied prompt; Flux.2 Pro and SD Pro added funny speech bubbles.
c4 · The Mustard Seed (photo). Near-identical photoreal results; 3D Diffusion drifts toward a Pixar face.
c4 · The Mustard Seed (photo). Near-identical photoreal results; 3D Diffusion drifts toward a Pixar face.
c5 · Title card. Spelling was right everywhere except Protogen; clean on DALL-E 3, Flux, GPT Image 1.5, Midjourney and SDXL.
c5 · Title card. Spelling was right everywhere except Protogen; clean on DALL-E 3, Flux, GPT Image 1.5, Midjourney and SDXL.
c6 · Infographic. Every engine got the labels right; GPT Image 1.5/2 and Protogen made it tiny.
c6 · Infographic. Every engine got the labels right; GPT Image 1.5/2 and Protogen made it tiny.

Calibration · 35 credits · 13 images

True cost per call, and do seeds work?

Question
What does each engine really charge, and does --seed reproduce an image?
Setup
Sequential runs with a balance check before and after each one, so no other job overlaps the measurement.
Verdict
Create costs 1 or 2 credits by engine family; edits cost 4–7. Seeds are ignored.
  • Create: fl_flux_pro2, fl_flux, gpt_image_15, gpt_image_2 = 1 · n_midjourney, n_sdxl (and by elimination the other n_*) = 2 · dalle3 = 19.
  • Edit: flux_pro2, flux_kontext, gpt_image_15 = 4 · gemini_flash, gpt_image_25 = 5 · seedream = 7.
  • Same seed twice: correlation 0.03 (Flux.2 Pro) and 0.17 (Midjourney), so unrelated images.
Same prompt and seed 7: two Flux.2 Pro runs (2, 3) and two Midjourney runs (6, 7) give different pictures.
Same prompt and seed 7: two Flux.2 Pro runs (2, 3) and two Midjourney runs (6, 7) give different pictures.

Round 2a · ≈170 credits · 36 images

Diagram or layout sketch → finished image

Question
Is image-to-image from a diagram better than text-to-image?
Setup
Six SVG sources built by hand (three labeled infographics, three rough blocking sketches), uploaded with airbrush upload and sent to all six edit engines with a "preserve the exact layout, keep text spelled exactly" prompt.
Verdict
Yes. Composition and labels come through intact, and the quality is publishable. Use gemini_flash or gpt_image_25 for 16:9.
  • Every engine kept all labels exactly, including small captions ("clinging causes it").
  • The edit engines also look alike (same hand lettering and rock painting), so the pooled backend applies here too.
  • Four of six engines turn a 16:9 input into a 1600² square.
Blocking sketch → gouache. Compare with Round 1 c1: the figures now stand exactly where they were drawn.
Blocking sketch → gouache. Compare with Round 1 c1: the figures now stand exactly where they were drawn.
Flat infographic → painted infographic, every label preserved.
Flat infographic → painted infographic, every label preserved.
The Middle Way: cliffs, path and signposts kept in place.
The Middle Way: cliffs, path and signposts kept in place.
The Four Ponds with a vignette per point of view, which text-only prompts never managed.
The Four Ponds with a vignette per point of view, which text-only prompts never managed.
Traffic layout → cartoon; the red swerving car lands where it was drawn.
Traffic layout → cartoon; the red swerving car lands where it was drawn.
The fear wheel as an illustrated loop.
The fear wheel as an illustrated loop.

Round 2c · ≈25 credits · 19 images

Prompt craft from the research guide

Question
Do research-backed prompts (the hero formula; tags for SD engines) improve results?
Setup
Four hero-formula prompts × the four strongest create engines, plus one tag-style prompt on three "SD" engines.
Verdict
The hero formula works everywhere: headline space, palette and no text on every run. Tag prompts made no difference, which fits the shared-backend theory.
  • Midjourney's four-pane pond (green, brown, underwater, mirror) is the best image of that metaphor so far.
  • GPT Image 1.5/2 turned the four sights into bold saffron ink-wash, a strong candidate for a house style.
Hero formula on four engines × four episode heroes. Left side kept empty for headlines.
Hero formula on four engines × four episode heroes. Left side kept empty for headlines.
Tag-style prompts on SD-named engines come back as modern photoreal images.
Tag-style prompts on SD-named engines come back as modern photoreal images.

Round 2d · ≈22 credits · 9 images

Can GPT Image 2.5 create from scratch?

Question
gpt_image_25 is listed as edit-only. Can it generate an image from a text prompt?
Setup
Same three prompts three ways: create --engine gpt_image_25, create --engine zz_fake_engine (control), and a gpt_image_25 edit on a blank 16:9 canvas.
Verdict
Yes, through the blank canvas. The create endpoint quietly falls back to a generic model for unknown engine names.
  • The create attempts and the fake engine both returned 1280×720 images from the same fallback model in 7–10 s, adding text nobody asked for.
  • The blank-canvas edit took 32–41 s and returned 2048×1152, with the best four-sights composition of the whole study.
Left: create gpt_image_25 · middle: fake engine control · right: real GPT Image 2.5 on a blank canvas.
Left: create gpt_image_25 · middle: fake engine control · right: real GPT Image 2.5 on a blank canvas.

Round 3 · 96 credits · 20 images

Photo references, multi-reference, and a style anchor

Question
Can a gallery photo steer composition, can two references be combined, and can one anchor image hold a style across new scenes?
Setup
Four Pexels photos from airbrush search restyled by three edit engines; one layout sketch + gate photo as a two-image edit; the gouache four-sights image used as Image 1 (style only) for three new scenes and one layout.
Verdict
Yes to all three. Photo → illustration keeps camera and composition; two references combine cleanly; the style anchor produced a coherent series.
  • The layout sketch + Jaipur gate photo gave the richest four-sights image of the study: real architecture, exact blocking.
  • Style anchor: the ascetics, the lane cutter and the row of doors all read as one illustrator's work.
  • Misses: the aerial traffic photo shrank the bubbles to dots, and a top-down layout restyled in gouache came out flat. Pick references whose camera angle suits the story.
Each row: source photo, then Flux.2 Pro edit (square), Nano Banana, GPT Image 2.5. Last row combines the layout sketch with the gate photo.
Each row: source photo, then Flux.2 Pro edit (square), Nano Banana, GPT Image 2.5. Last row combines the layout sketch with the gate photo.
Style anchor (1) and new scenes painted from it by Nano Banana and GPT Image 2.5.
Style anchor (1) and new scenes painted from it by Nano Banana and GPT Image 2.5.

Production · 90 credits · 18 images

Episode image sets from the style anchor

Question
Can the recommended pipeline fill every story beat of three episodes in one consistent style?
Setup
Seven beats generated with both 16:9 engines from the gouache anchor (best one kept), then four more with GPT Image 2.5 alone. Diagrams and photo-based scenes reused from the research rounds.
Verdict
Yes. 20–23 credits per episode for a full, consistent set of 7–8 images.
  • Picks favored true 16:9 output and clean headline space; both engines won some pairs.
  • Every episode set lives in episodes/NN/images/ with a manifest recording the source engine, prompt and hosted URL.
Seven beats, each from both engines.
Seven beats, each from both engines.
Four extra beats from GPT Image 2.5.
Four extra beats from GPT Image 2.5.

VideosThree styles, with and without images

Six Haiku agents (model haiku, effort high) each built one style in its own Remotion project. Every style is built twice, with and without the airbrush images, to measure what the illustrations add. All six share one data pipeline: the script, then the voice, then Spokenly word timings, then scene timing. Swapping in the ElevenLabs voice later re-renders in sync without code changes. Narration uses ElevenLabs Eleven v4 ("our fastest and most emotive model", available on the account alongside v3), generated with sag. Each style has its own premade voice. v4 performs the inline tags ([sarcastic], [laughs], [pause]) instead of reading them aloud, confirmed by transcribing test clips.

StyleFrameLook
Ink & Paper16:9Contemplative storybook: full-bleed gouache with slow push-ins, serif type, a quiet caption band. No-image version: rice paper, a self-drawing ensō circle, ink-stroke glyphs.
Punk Zine9:16The Bad Buddha's attitude: torn-paper cutouts, halftone, huge condensed caps, word-by-word captions, stickers. No-image version: pure kinetic typography and marker scribbles.
Diagram Explainer16:9Clear motion graphics: image cards with bullet notes that build as phrases are spoken, a chapter progress bar. No-image version: bespoke code-drawn diagrams (four ponds, fear wheel, middle way).

Voices

StyleVoiceCharacterSetting
Ink & PaperAdam LiDeep, steady and calmeleven_v4_turbo
Punk ZineCallumHusky trickstereleven_v4
Diagram ExplainerAliceClear, engaging educatoreleven_v4

Skills written for the agents

  1. The images matter a lot for the cinematic style and hardly at all for the typographic one.

    Ink & Paper without images falls back to headlines on empty paper with thin ink glyphs, and the storybook feel goes with the pictures. Punk Zine without images is as strong as with them: ransom-note typography, giant list numbers and bold icons carry it alone.

  2. Code beats AI images for diagrams; AI images beat code for stories.

    The no-image Diagram Explainer drew clearer concept diagrams (the four sights as figures, numbered key-point cards, the fear loop) than the AI diagram images, which are small and busy once overlaid. Its story scenes ("Starving for answers", "The prince in the palace") sit nearly empty without a picture. The best version would mix the two: code diagrams for concepts, illustrations for stories.

  3. Haiku at high effort produced six working designs from one playbook.

    Each agent read both skills, built its style in 42–83 tool calls (10–14 minutes), typechecked, checked 10–30 stills, and fixed real layout bugs (overflowing stickers, wrapped labels, empty bottom halves). They needed one correction from the coordinator: parallel renders collided on ports. All final renders ran one at a time on dedicated ports.

  4. Timing never drifted when the voice changed.

    All six projects were built against a temporary macOS voice, then re-rendered with ElevenLabs v4 without code changes, because every animation reads its timing from the word-level data. Durations: 117–128 s per video.

  5. Judge the voices by ear.

    Claude can check timing and transcripts but can't hear delivery. The three v4 voices were chosen from ElevenLabs' own descriptions and verified only for tag handling (no tags read aloud) and pacing (116–127 s for the same script). Which voice fits Bad Buddhist Radio best is your call.

16:9 · George

Ink & Paper

Chosen format. Re-scripted to cover every key point and story from each episode, with 15–17 scenes and 3.5–4.3 minutes per video. There's no show name and no road joke; each video ends on its own closing line. Narrated by Adam Li on Eleven v4 Turbo, which bills half the characters of full v4, with vocal-only audio tags per ElevenLabs' v4 prompting guide. Five new gouache images cover the added beats. The no-image version on the right is the earlier, shorter cut.

With airbrush images

Episode 1 · 8 MB preview
Episode 2 · 11 MB preview
Episode 3 · 9 MB preview

Without images (all visuals drawn in code)

Episode 1 · 2 MB preview
Episode 2 · 2 MB preview
Episode 3 · 2 MB preview

9:16 · Callum

Punk Zine

Vertical 9:16 for Shorts and Reels, voiced by Callum ("husky trickster"). With images, tilted torn-paper cutouts of the gouache art slam in under saffron tape headlines. Without images, the ransom-note kinetic type, giant numbers (1 OLD MAN, 2 SICK MAN…) and bold icons are just as punchy, and arguably more on-brand. Both leave the lower third a little empty.

With airbrush images

Episode 1 · 6 MB preview
Episode 2 · 7 MB preview
Episode 3 · 6 MB preview

Without images (all visuals drawn in code)

Episode 1 · 4 MB preview
Episode 2 · 4 MB preview
Episode 3 · 4 MB preview

16:9 · Alice

Diagram Explainer

Voiced by Alice ("clear, engaging educator"). With images it gets a steady rhythm of an image card plus key points that build as they're spoken, though the images sit small. Without images it gets bespoke animated diagrams for the eight concept scenes, which explain better than the AI images, but story scenes are nearly empty. Combining both would be the strongest version.

With airbrush images

Episode 1 · 4 MB preview
Episode 2 · 5 MB preview
Episode 3 · 4 MB preview

Without images (all visuals drawn in code)

Episode 1 · 3 MB preview
Episode 2 · 3 MB preview
Episode 3 · 3 MB preview

EpisodesChapters, ideas, stories, posts

Episode 1 · 08:04 · host Ryu Cope

Just Some Dude Who Woke Up

Who the Buddha really was, and why Bad Buddhism strips him back down to size

Ryu Cope opens Bad Buddhist Radio with the Buddha's origin story told without the incense: a sheltered prince, four shocking sights, a near-fatal fast, and a man who insisted he was not a god, only awake. From there he lays out what Buddhism is (here-and-now, practical, moderate) and what makes his version 'bad': no karma as cosmic bookkeeping, no reincarnation, no superhuman Buddha. If an ordinary man could wake up, so can you, and that means you have no excuse.

Chapters

  1. 00:00Putting the Boot Back in Buddhism
  2. 00:28The Prince Who Never Saw Suffering
  3. 00:56The Four Sights
  4. 01:37Starving for Answers
  5. 02:35"I'm Not a God, I'm Awake"
  6. 03:02Here and Now, Practical, Moderate
  7. 04:50What Makes It Bad
  8. 05:16Own It: No Karma, No Excuses
  9. 05:46The Myths Added Later
  10. 06:49Just Some Dude
  11. 07:18Good News, Bad News
  12. 07:50Buy Him a Beer

Key ideas

  • Everybody gets old, everybody gets sick, everybody dies. The Buddha's whole search starts from refusing to look away from that.
  • Starving yourself for enlightenment fails for a simple reason: if you're dead, you can't find any answers. The middle way begins with lunch.
  • 'Buddha' is a job description, not a title: the awakened one. 'I'm not a god, I'm awake.'
  • Buddhism cares about this life, right now; it is practical; and it walks the hard line between extremes.
  • Bad Buddhism drops karma and rebirth as reward-and-punishment. Do good because it's right now; when you do bad, own it.
  • The Buddha was an ordinary man. Good news: anyone can wake up. Bad news: you have no excuse, and 2,500 years of teaching to help you.

Stories to emphasize

  • The Four Sights. A sheltered prince sneaks out and meets an old man, a sick man, a dead man and a content holy man.
  • The Starving Ascetics. He nearly fasts to death with the ascetics, then realizes the mind needs the body and goes to eat.
  • Are You a God?. Old friends don't recognize him. One asks if he is a god. 'No. I'm awake.'
Hero · `gpt_image_25`
Hero · gpt_image_25
The Four Sights (layout sketch + Jaipur gate photo) · `gemini_flash`
The Four Sights (layout sketch + Jaipur gate photo) · gemini_flash
Starving for answers · `gpt_image_25`
Starving for answers · gpt_image_25
“I’m awake” (from a gallery photo) · `gpt_image_25`
“I’m awake” (from a gallery photo) · gpt_image_25
The Middle Way (from an SVG diagram) · `gemini_flash`
The Middle Way (from an SVG diagram) · gemini_flash
Own it · `gpt_image_25`
Own it · gpt_image_25
Just some dude · `gpt_image_25`
Just some dude · gpt_image_25
Title card (GPT Image 1.5) · `gpt_image_15`
Title card (GPT Image 1.5) · gpt_image_15
Read the blog post

Bad Buddhist Radio, Episode 1 · Ryu Cope · 8 minutes

In 2006, Ryu Cope opened a podcast with a promise to put "the boot back in Buddhism." His first lesson goes after the man at the center of it all. Not to knock him down, but to take him off the pedestal and put him back on the road, where he started.

A prince who had never seen a bad day

Siddhartha Gautama was born a prince in what is now Nepal, around 563 BCE. His father wanted him to never suffer, which in practice meant keeping him locked inside the palace walls. It worked until it didn't. One day the prince talked a servant into sneaking him out, and on the road he saw four things he had never seen before: an old man, a sick man, a dead man, and a holy man.

The servant didn't soften it. Everybody gets old. Everybody gets sick. Everybody dies. Period.

The first three sights terrified him. The fourth puzzled him: the holy man seemed content, knowing everything the prince had just learned. That contentment became the question Siddhartha would spend his life on.

Starving for answers

He left the palace, his wife, his child and his inheritance, and went looking. He tried a bit of everything before joining a group of ascetics who fasted and punished their bodies in the hope of spiritual breakthrough. He starved himself almost to death before he noticed the flaw in the plan.

"If you're dead you can't find any kind of answers."

So he got up and ate. Ryu treats this as the real beginning of Buddhism. The mind needs the body, the body needs the mind, and neither works if you destroy the other. The middle way starts with lunch.

"I'm not a god, I'm awake"

When he finally understood what he was looking for, he went back to the friends he had studied with. He had changed so much that they didn't recognize him. One asked whether he was a god.

"No, I'm not a god, I'm awake."

That's more or less what Buddha means: the awakened one. It's a description, not a title.

Three things Buddhism actually is

Ryu boils the teaching down to three points.

  • Here and now. Buddhism isn't concerned with gods, spirits or afterlives. It cares about this life, this moment.
  • Practical. The Buddha lived in the ordinary world and taught about the things ordinary people struggle with.
  • Moderate. The middle way runs between handing your life to dogma and doing whatever you feel like. Walking that line is the hard part, because following orders is easy and thinking for yourself isn't.

What makes it "bad"

Here's where Bad Buddhism parts ways with most modern Buddhism. Ryu argues that the tradition has been cluttered with "mystical nonsense and metaphysical mumbo jumbo," and he strips it out.

Karma and reincarnation go first. As Ryu sees them, they're a cosmic system of reward and punishment that pulls your attention away from the present. Do good because it's the right thing to do now, not for points in this life or the next. And when you do something bad:

"If you do something bad, own it."

Not karma, not evil spirits, not temptation. You.

Next to go are the later legends: the Buddha choosing to delay nirvana out of infinite compassion, his godlike birth, his place in a lineage of cosmic Buddhas. Ryu's verdict is short: added later, by people who wanted the Buddha to compete with the gods.

Good news, bad news

What's left after all that stripping is the most radical claim in the episode. The Buddha was an ordinary man. "He was just some dude." He figured out how to deal with life on his own, without a god dictating to him.

That cuts both ways. The good news: if an ordinary person could wake up, anybody can. The bad news: you have no excuse. And unlike the Buddha, you have twenty-five centuries of teaching to help.

Ryu's sign-off sets the tone for everything that follows: if you meet the Bad Buddha on the side of the road, buy him a beer.

Chapters

  • 00:00 Putting the Boot Back in Buddhism
  • 00:28 The Prince Who Never Saw Suffering
  • 00:56 The Four Sights
  • 01:37 Starving for Answers
  • 02:35 "I'm Not a God, I'm Awake"
  • 03:02 Here and Now, Practical, Moderate
  • 04:50 What Makes It Bad
  • 05:16 Own It: No Karma, No Excuses
  • 05:46 The Myths Added Later
  • 06:49 Just Some Dude
  • 07:18 Good News, Bad News
  • 07:50 Buy Him a Beer

Episode 2 · 15:04 · host Ryu Cope

Get Out of Your Bubble

Enlightenment without the mysticism: insight, empathy, and the guy who cut you off in traffic

Why bother with enlightenment at all? Ryu Cope takes the dictionary route: enlightenment is simply possessing insight. Nobody can know everything, so 'one with the universe' is out. What's left is the ability to step outside your own bubble of perceptions and see through someone else's eyes, or see the world as it truly is. He tells the story of the four ponds, explains how uncertainty breeds fear and fear breeds anger, and replays a near-miss in traffic twice: once asleep, once awake.

Chapters

  1. 00:00Welcome Back
  2. 00:32So What's the Big Deal?
  3. 01:24Dictionary Enlightenment
  4. 02:35Nobody Knows Everything
  5. 03:16The Four Ponds
  6. 05:09Living in Bubbles
  7. 05:36Walk a Mile in Their Moccasins
  8. 06:29Why Most People Don't Bother
  9. 07:56The Smart Man and the Wise Man
  10. 08:49Uncertainty, Fear, Anger
  11. 09:48The Lane Cutter
  12. 12:26A Thousand Little Clues
  13. 13:27The Wheel Keeps Spinning
  14. 13:53Natural, Not Supernatural

Key ideas

  • Strip off the mysticism and 'enlightenment' just means possessing insight. 'Enlighten me' is something you say every day.
  • No human can know everything, so 'one with the universe' is off the table. Enlightenment is empathy: seeing through someone else's eyes, and seeing the world as it truly is.
  • We all live in bubbles built from our own perceptions, memories and experiences. Nobody chose it. It just is.
  • The smart man learns from his own mistakes, the wise man learns from others'. The awake man learns before the mistake happens.
  • Uncertainty creates fear; fear turns into anger, defensiveness and self-righteousness. Insight breaks the chain.
  • Enlightenment is ongoing maintenance, like breathing or eating. It's natural, not supernatural.

Stories to emphasize

  • The Four Ponds. A monk asks his student how many ponds there are. One? No: the monk's pond, the student's pond, the koi's pond, and the pond as it truly is.
  • The Lane Cutter. A driver realizes too late he's in the wrong lane and swerves; you, lost in your own bubble, honk and curse. Both ruin your day. One moment of awareness on either side would have prevented it.
Hero · `gpt_image_25`
Hero · gpt_image_25
The Four Ponds (from an SVG diagram) · `gemini_flash`
The Four Ponds (from an SVG diagram) · gemini_flash
Monk and student (from a gallery photo) · `gpt_image_25`
Monk and student (from a gallery photo) · gpt_image_25
Everybody lives in a bubble · `gemini_flash`
Everybody lives in a bubble · gemini_flash
The wheel that keeps spinning (from an SVG diagram) · `gemini_flash`
The wheel that keeps spinning (from an SVG diagram) · gemini_flash
The lane cutter · `gemini_flash`
The lane cutter · gemini_flash
Smart, wise, awake · `gpt_image_25`
Smart, wise, awake · gpt_image_25
Read the blog post

Bad Buddhist Radio, Episode 2 · Ryu Cope · 15 minutes

Why do people meditate, chant and study for years just to reach enlightenment? In the second episode of Bad Buddhist Radio, Ryu Cope asks what enlightenment actually is. His answer has nothing to do with dissolving into the universe, and a lot to do with the guy who cut you off this morning.

Enlightenment, by the dictionary

Ryu starts by throwing out "all the BS about endless suffering and rebirth" and just reading the words. Nirvana, as most Buddhists define it, is a state of perfect wisdom and infinite compassion: the "one with everything" thing. Enlightenment, stripped of metaphysics, is the state of having been given knowledge or insight. You use the word yourself whenever you say "enlighten me."

So if enlightenment is having insight, nirvana would be having all insight. Ryu's response: no human being can know everything, so that version is off the table. What's left is something smaller and more useful. Enlightenment is the ability to step outside your own little world and see through someone else's eyes, or better still, to see the world as it truly is.

The four ponds

To show what that means, Ryu tells an old story. A monk sits with his student beside a koi pond and asks how many ponds there are. One, says the student. Wrong, says the monk. There are four.

"I didn't ask you how many you saw, I asked you how many there were."

The monk explains. The pond he experiences isn't the pond the student experiences. Maybe one sees the water as greener, the other as browner; maybe one can smell the lilies. And neither of their ponds is the pond the koi lives in. To the humans, the water is something pretty to look at. To the koi, it's life.

The student works it out: that's only three ponds. The monk smiles.

"You're forgetting the most important pond of all, the pond as it truly is."

Everybody lives in a bubble

The point is that all of us wander around inside bubbles made of our own perceptions, memories and experiences. Nobody chooses this. It's how being a person works. When somebody does something, our first and often only question is how it affects us.

Enlightenment, in Ryu's telling, is the ability to leave your bubble for a while and look into someone else's. When you do, you understand people a little better. Understanding leads to empathy, and empathy leads to treating people better. "Oh no, what's that? Is that that whole compassion thing?"

Why most people don't bother

Anybody can do this, so why doesn't everybody? Ryu gives three reasons: people don't know how, it's hard work, and it never stays done. That last one is the real catch. Enlightenment isn't a certificate you earn once. It's maintenance, like breathing, eating or sleeping. You never take one drink and call yourself hydrated for life.

The payoff is worth it, though. Most problems trace back to a lack of insight. His version of an old saying puts it neatly: the smart man learns from his own mistakes, the wise man learns from the mistakes of others, and the awake man learns before the mistake happens at all.

Uncertainty, fear, anger

Ryu adds a piece of psychology. Uncertainty creates fear. Fear breeds anger, defensiveness and self-righteousness, or worse, a numbness that leaves people mindless and joyless. Insight breaks that chain, because once you understand why something happened, you can do something about it, even if that something is only empathy.

The lane cutter

Then comes the story the episode is built around. You're driving. A guy suddenly swerves across your lane because he just realized he needs to turn right. He doesn't hit you, but he could have. So you honk, flip him off and call him a few names. Now he's mad at you. You both spend the rest of the day in a bad mood.

Replay it with one person awake. If the other driver had been paying attention, he'd have known where he was, or he'd have realized missing a turn is no big deal and taken the next one. Instead, a split second of not knowing scared him, the fear made him defensive, and he cut across a lane because in that moment only his own bubble mattered.

And if you had been awake, you might have seen a thousand little clues: him slowing at every intersection, squinting at street signs, talking on his phone.

"It ain't ESP. It's called being aware."

Instead the cycle keeps turning. Fear, anger, back into the bubble, round and round like the battery bunny. Reacting is easy. Seeing outside yourself is work, and you have to do it again next time.

Natural, not supernatural

Ryu's closing argument is that enlightenment isn't reserved for old monks in robes who have pondered riddles for decades. It's a natural state of understanding and insight that anyone can reach by paying attention and living in the moment.

"Insight is natural, people, not supernatural."

In other words, try not to be a selfish, self-centered moron. That's the lesson.

Chapters

  • 00:00 Welcome Back
  • 00:32 So What's the Big Deal?
  • 01:24 Dictionary Enlightenment
  • 02:35 Nobody Knows Everything
  • 03:16 The Four Ponds
  • 05:09 Living in Bubbles
  • 05:36 Walk a Mile in Their Moccasins
  • 06:29 Why Most People Don't Bother
  • 07:56 The Smart Man and the Wise Man
  • 08:49 Uncertainty, Fear, Anger
  • 09:48 The Lane Cutter
  • 12:26 A Thousand Little Clues
  • 13:27 The Wheel Keeps Spinning
  • 13:53 Natural, Not Supernatural

Episode 3 · 10:39 · host Ryu Cope

All Life Includes Suffering

The First Noble Truth, minus the angsty goth nonsense

Ryu Cope maps out the Four Noble Truths and the Eightfold Path, admits he can't stand how preachy the Truths sound, then takes on the first one. 'All life is suffering' has produced two myths: that Buddhists are depressed nihilists, and that life is literally nothing but misery. The Bad Buddhist reading is simpler: all life includes suffering. You're not the only one who suffers, you're not the only one facing your problem, and your suffering is a matter of perspective. Then comes the parable of the mustard seed.

Chapters

  1. 00:00Welcome and Shout-Outs
  2. 00:25The 12-Step Program
  3. 00:58The Four Noble Truths in Brief
  4. 01:27The Eightfold Path in Brief
  5. 01:59Why I Don't Like the Noble Truths
  6. 02:58"All Life Is Suffering"?
  7. 03:29Myth One: Depressed Nihilists
  8. 03:57Myth Two: Take It Literally
  9. 04:48Life Includes Suffering
  10. 05:45The Wolf and the Bunny
  11. 06:17You're Not the Only One
  12. 06:51The Mustard Seed
  13. 08:21Suffering Is Perspective
  14. 09:16Cowboy Up
  15. 10:20He Could Use a Ride

Key ideas

  • The Four Noble Truths: suffering, attachment, escape, and the middle way, which is the Eightfold Path.
  • Even a Buddhist can find the Truths preachy: problem, cause, good news, follow our steps.
  • 'All life is suffering' breeds two myths: Buddhists as depressed nihilists, and life as literal non-stop misery.
  • The Bad Buddhist version: all life INCLUDES suffering. Life also includes joy, love and pleasure, and denying either is denying reality.
  • For the wolf to eat, a bunny was lunch. For your green light, someone sits at a red. Suffering is built into the system.
  • Three keys: you're not the only one who suffers; you're not the only one facing your problem; and suffering is a matter of perspective.

Stories to emphasize

  • The Mustard Seed. A grieving mother carries her dead child to the Buddha. He asks for a mustard seed from a house that has never known death. Door after door, she learns no such house exists, and lets her child go.
  • The Car Seats. A friend whines that he can't afford new seats for his car. What about the people who can't afford a car, or food?
  • No Shoes, No Feet. 'I cried because I had no shoes till I saw the man who had no feet.'
Hero · `gemini_flash`
Hero · gemini_flash
The Four Noble Truths (from an SVG diagram) · `gemini_flash`
The Four Noble Truths (from an SVG diagram) · gemini_flash
Angsty goth nonsense · `gpt_image_25`
Angsty goth nonsense · gpt_image_25
All life includes suffering · `gpt_image_25`
All life includes suffering · gpt_image_25
The wolf and the bunny · `gemini_flash`
The wolf and the bunny · gemini_flash
Every door has known death · `gpt_image_25`
Every door has known death · gpt_image_25
Perspective · `gpt_image_25`
Perspective · gpt_image_25
Title card (GPT Image 1.5) · `gpt_image_15`
Title card (GPT Image 1.5) · gpt_image_15
Read the blog post

Bad Buddhist Radio, Episode 3 · Ryu Cope · 11 minutes

The First Noble Truth is usually translated as "all life is suffering." In episode three of Bad Buddhist Radio, Ryu Cope argues that this single word, is, has done more damage to how people see Buddhism than anything else. He changes one word and the whole teaching makes sense.

The map: four truths and eight steps

Ryu calls it "the Buddha's 12-step program." The Four Noble Truths are what the Buddha realized on the night of his enlightenment:

  • the truth of suffering
  • the truth of attachment, the cause of suffering
  • the truth of escape, that suffering can end
  • the truth of the middle way, the path that ends it

The middle way is the Eightfold Path: right view, right intention, right speech, right action, right livelihood, right effort, right mindfulness and right concentration. There are "a bajillion" versions of both lists, he admits. Push comes to shove, they all mean about the same thing.

Why a Buddhist doesn't like the Noble Truths

Then comes a surprise. Ryu tells his listeners he doesn't care for the Four Noble Truths at all. They sound too much like the street-corner preacher: you're broken, here's why, but there's good news, just follow our steps. Problem, cause, promise, method. He concedes that every religion has that shape. He still doesn't have to like it.

Two myths from one word

"All life is suffering" leads to two common myths, Ryu says.

Myth one: Buddhists are depressed nihilists. If life is nothing but misery and you'll be reborn anyway, why bother living? He hears this one a lot, living in Middle America.

Myth two: the truth is meant 100% literally. You suffer from birth to death, physically, mentally and spiritually, and you'll never be happy until you reach nirvana. A lot of modern Buddhists talk this way, which is where myth one comes from.

His response is blunt:

"All life is not suffering."

Believing it is "a bunch of angsty goth nonsense." Life includes joy, love, happiness and pleasure, and denying that is denying reality, which the Buddha never taught.

One word changes everything

The Bad Buddhist version of the First Noble Truth:

"All life includes suffering. Period. Easy as that."

You're going to suffer. Life can really bite sometimes, and it can also be really awesome. Suffering is built into how the world works, even when it isn't yours. For the wolf to be happy with a full belly, a bunny had to be lunch. For you to get a green light, someone else has to sit at a red one.

That isn't pessimism, Ryu insists. It's realism.

Three keys to living with it

A: You're not the only one who has ever suffered. Humans have been around for tens of thousands of years. Gazillions came before you.

B: You're not the only one facing your particular problem. Unless you have a condition that exists only in you, other people have been sick, lost, grieving, out of work or on the street, and some of them are there right now, alongside you.

To show it, Ryu tells the parable of the mustard seed. A mother's child dies, and she carries his body from place to place looking for someone who can bring him back. She takes him to the Buddha. The Buddha can't perform miracles, but he can help her understand. He promises to revive the boy if she brings him a handful of mustard seeds from a house that has never known death.

She goes door to door. Every family has lost someone. By the end of the street she understands that death is part of being human, and she lets her child go.

"Sorry to say it, but you ain't special."

C: Suffering is a matter of perspective. Ryu remembers a friend who complained he couldn't afford new seats for his car. What about the people who can't afford a car, or food? As the old saying goes, I cried because I had no shoes till I saw the man who had no feet. It's harsh, he admits, but reality is harsh if you don't face up to it.

Cowboy up

That's what being awake comes down to: facing reality so you can stare it down and be happy anyway. Once you really understand the First Noble Truth, your suffering stops looking special. It's par for the course, as natural as breathing, as uncontrollable as the color of the sky. You already accept those.

So, in Ryu's words, cowboy up, and drop the idea that you or anyone else should never have to suffer. Next time: the Second Noble Truth, and why you're suffering because of your attachment to, well, not suffering.

Chapters

  • 00:00 Welcome and Shout-Outs
  • 00:25 The 12-Step Program
  • 00:58 The Four Noble Truths in Brief
  • 01:27 The Eightfold Path in Brief
  • 01:59 Why I Don't Like the Noble Truths
  • 02:58 "All Life Is Suffering"?
  • 03:29 Myth One: Depressed Nihilists
  • 03:57 Myth Two: Take It Literally
  • 04:48 Life Includes Suffering
  • 05:45 The Wolf and the Bunny
  • 06:17 You're Not the Only One
  • 06:51 The Mustard Seed
  • 08:21 Suffering Is Perspective
  • 09:16 Cowboy Up
  • 10:20 He Could Use a Ride

CostsCredit ledger

RoundImagesCreditsNote
Round 1 · engine screen86182Includes 38 credits for two DALL-E 3 images
Calibration · creates811Sequential, balance-checked
Calibration · edits524Sequential, balance-checked
Round 2a · diagram → image (rest)31140Estimated from measured per-engine prices
Round 2c · prompt craft1925
Round 2d · gpt_image_25 test1023
Round 3 · photo reference + style anchor209610 photo/multi-reference edits, 10 style-anchor edits
Production · episode images1890Style-anchored story images for scenes the research set didn't cover
Production · extra story images525Leaving the palace, the legends, the lost driver, the eightfold path, four steps

LogWhat happened, in order

  1. 21:02Spokenly transcribed episodes 1–3 in about 30 s; speaker labels need Spokenly Pro, so the transcripts have none (it's one host).
  2. 21:05Chapters anchored to exact phrases and timed from word-level timestamps.
  3. 21:08Round 1 launched: 86 images across 15 engines.
  4. 21:20Shared-backend pattern spotted; calibration confirms prices and that seeds are ignored.
  5. 21:30Diagrams uploaded; Round 2 edits and prompt-craft runs.
  6. 21:55GPT Image 2.5 works as a creator through a blank-canvas edit.
  7. 22:10Round 3: photo references, a two-image edit and the style anchor all work. Research phase complete; production images next.
  8. 22:40Blog posts written; image sets assembled (20–23 credits per episode).
  9. 22:55Remotion starter verified; skills written; six Haiku agents started (three styles × with/without images).
  10. 23:20sag works through the interactive shell profile; Eleven v4 found. Nine narrations generated and timed (George, Callum, Alice).
  11. 23:40All six agents finished. 18 final videos rendered with v4 voices, one at a time on dedicated ports (≈2 min each).
  12. 06:20Ink & Paper chosen. Scripts rewritten (all key points and stories, no show name or closing joke); Adam Li on Eleven v4 Turbo; 3 videos re-rendered (216 s, 262 s, 210 s).