I’m a permaculture farmer. I want to create food-producing ecosystems. My hope is a world with easy access to a cuisine that nourishes consumers, supports producers, and leaves the Earth joyously habitable.

Permaculturists, natural farmers, plantsmen, and foodies share this ambition. I believe this group of green thumbs, stock-folk, and food champions is falling to tribalism, forgetting that rescuing the globe requires saving all of its inhabitants, even those who adore cheap burgers and Coke. We're digging foxholes and turning folks who disagree with us or don't understand into monsters.

Take Dr. Daphne Miller's comments at the end of her Slow Money Journal interview:

“Americans are going to fall into two camps when all is said and done: People who buy cheap goods, regardless of quality, versus people who are willing and able to pay for things that are made with integrity. We are seeing the limits of the “buying cheap crap” approach.”

This is one of the most judgmental things I've read outside the Bible. Consequences:

• People who purchase inexpensive things (food) are ignorant buffoons who prefer to choose fair trade coffee over fuel as long as the price is correct.

• It all depends on your WILL to buy quality or cheaply. Both those who are WILLING and those who ARE NOT exist. And able, too.

• People who are unwilling and unable are purchasing garbage. You're giving your kids bad food. Both the Earth and you are being destroyed by your actions. Your camp is the wrong one. You’re garbage! Disgrace to you.

Dr. Miller didn't say it, but words are worthless until interpreted. This interpretation depends on the interpreter's economic, racial, political, religious, family, and personal history. Complementary language insults another. Imagine how that Brown/Harvard M.D.'s comment sounds to a low-income household with no savings.

Dr. Miller's comment reflects the echo chamber into which nearly all clean food advocates speak. It asks easy questions and accepts non-solutions like raising food prices and eating less meat. People like me have cultivated an insular world unencumbered by challenges beyond the margins. We may disagree about technical details in rotationally-grazing livestock, but we short circuit when asked how our system could supply half the global beef demand. Most people have never seriously considered this question. We're so loved and affirmed that challenging ourselves doesn't seem necessary. Were generals insisting we don't need to study the terrain because God is on our side?

“Yes, the $8/lb ground beef is produced the way it should be. Yes, it’s good for my body. Yes it’s good for the Earth. But it’s eight freaking dollars, and my kid needs braces and protein. Bye Felicia, we’re going to McDonald’s.”

-Bobby Q. Homemaker

Funny clean foodies. People don't pay enough for food; they should value it more. Turn the concept of buying food with integrity into a wedge and drive it into the heart of America, dividing the willing and unwilling.

We go apeshit if you call our products high-end.

I've heard all sorts of gaslighting to defend a $10/lb pork chop as accessible (things I’ve definitely said in the past):

• At Whole Foods, it costs more.

• The steak at the supermarket is overly affordable.

• Pay me immediately or the doctor gets paid later.

I spoke with Timbercreek Market and Local Food Hub in front of 60 people. We were asked about local food availability.

They came to me last, after my co-panelists gave the same responses I would have given two years before.

I grumbled, "Our food is inaccessible." Nope. It's beyond the wallets of nearly everyone, and it's the biggest problem with sustainable food systems. We're criminally unserious about being leaders in sustainability until we propose solutions beyond economic relativism, wishful thinking, and insisting that vulnerable, distracted people do all the heavy lifting of finding a way to afford our food. And until we talk about solutions, all this preserve the world? False.

The room fell silent as if I'd revealed a terrible secret. Long, thunderous applause followed my other remarks. But I’m probably not getting invited back to any VNRLI events.

I make pricey cuisine. It’s high-end. I have customers who really have to stretch to get it, and they let me know it. They're forgoing other creature comforts to help me make a living and keep the Earth of my grandmothers alive, and they're doing it as an act of love. They believe in us and our work.

I remember it when I'm up to my shoulders in frigid water, when my vehicle stinks of four types of shit, when I come home covered in blood and mud, when I'm hauling water in 100-degree heat, when I'm herding pigs in a rainstorm and dodging lightning bolts to close the chickens. I'm reminded I'm not alone. Their enthusiasm is worth more than money; it helps me make a life and a living. I won't label that gift less than it is to make my meal seem more accessible.

Not everyone can sacrifice.

Let's not pretend we want to go back to peasant fare, despite our nostalgia. Industrial food has leveled what rich and poor eat. How food is cooked will be the largest difference between what you and a billionaire eat. Rich and poor have access to chicken, pork, and beef. You might be shocked how recently that wasn't the case. This abundance, particularly of animal protein, has helped vulnerable individuals.

Industrial food causes environmental damage, chronic disease, and distribution inequities. Clean food promotes non-industrial, artisan farming. This creates a higher-quality, more expensive product than the competition; we respond with aggressive marketing and the "people need to value food more" shtick geared at consumers who can spend the extra money.

The guy who is NOT able is rendered invisible by clean food's elitist marketing, which is bizarre given a.) clean food insists it's trying to save the world, yet b.) MOST PEOPLE IN THE WORLD ARE THAT GUY. No one can help him except feel-good charities. That's crazy.

Also wrong: a foodie telling a kid he can't eat a 99-cent fast food hamburger because it lacks integrity. Telling him how easy it is to save his ducketts and maybe have a grass-fed house burger at the end of the month as a reward, but in the meantime get your protein from canned beans you can't bake because you don't have a stove and, even if you did, your mom works two jobs and moonlights as an Uber driver so she doesn't have time to heat that shitup anyway.

A wealthy person's attitude toward the poor is indecent. It's 18th-century Versailles.

Human rights include access to nutritious food without social or environmental costs. As a food-forest-loving permaculture farmer, I no longer balk at the concept of cultured beef and hydroponics. My food is out of reach for many people, but access to decent food shouldn't be. Cultures and hydroponics could scale to meet the clean food affordability gap without externalities. If technology can deliver great, affordable beef without environmental negative effects, I can't reject it because it's new, unusual, or might endanger my business.

Why is your farm needed if cultured beef and hydroponics can feed the world? Permaculture food forests with trees, perennial plants, and animals are crucial to economically successful environmental protection. No matter how advanced technology gets, we still need clean air, water, soil, greenspace, and food.

Clean Food cultivated in/on live soil, minimally processed, and eaten close to harvest is part of the answer, not THE solution. Clean food advocates must recognize the conflicts at the intersection of environmental, social, and economic sustainability, the disproportionate effects of those conflicts on the poor and lower-middle classes, and the immorality and impracticality of insisting vulnerable people address those conflicts on their own and judging them if they don't.

Our clients, relatives, friends, and communities need an honest assessment of our role in a sustainable future. If we're serious about preserving the world, we owe honesty to non-customers. We owe our goal and sanity to honesty. Future health and happiness of the world left to the average person's pocketbook and long-term moral considerations is a dismal proposition with few parallels.

Let's make soil and grow food. Let the lab folks do their thing. We're all interdependent.

Research on preventing disease, extending life, and caring for your body from the inside out

Grapefruit and pomegranates aren't on the list, so ignore that. Mostly, I enjoyed the visual, but those fruits are healthful, too.

15 (or 17 if you consider the photo) different foods a day sounds like a lot. If you're not used to it  — it is.

These lists don't aim for perfection. Instead, use this article and the science below to eat more of these foods. If you can eat 5 foods one day and 5 the next, you're doing well. This list should be customized to your requirements and preferences.

“Every time you eat or drink, you are either feeding disease or fighting it” -Heather Morgan.

The 15 Foods That You Should Consume Daily and Why:

1. Dark/Red Berries

(blueberries, blackberries, acai, goji, cherries, strawberries, raspberries)

The 2010 Global Burden of Disease Study is the greatest definitive analysis of death and disease risk factors in history. They found the primary cause of both death, disability, and disease inside the United States was diet.

Not eating enough fruit, and specifically berries, was one of the best predictors of disease (1).

What's special about berries? It's their color! Berries have the most antioxidants of any fruit, second only to spices. The American Cancer Society found that those who ate the most berries were less likely to die of cardiovascular disease.

2. Beans

Soybeans, black beans, kidney beans, lentils, split peas, chickpeas.

Beans are one of the most important predictors of survival in older people, according to global research (2).

For every 20 grams (2 tablespoons) of beans consumed daily, the risk of death is reduced by 8%.

Soybeans and soy foods are high in phytoestrogen, which reduces breast and prostate cancer risks. Phytoestrogen blocks the receptors' access to true estrogen, mitigating the effects of weight gain, dairy (high in estrogen), and hormonal fluctuations (3).

3. Nuts

(almonds, walnuts, pecans, pistachios, Brazil nuts, cashews, hazelnuts, macadamia nuts)

Eating a handful of nuts every day reduces the risk of chronic diseases like heart disease and diabetes. Nuts also reduce oxidation, blood sugar, and LDL (bad) cholesterol, improving arterial function (4).

Despite their high-fat content, studies have linked daily nut consumption to a slimmer waistline and a lower risk of obesity (5).

4. Flaxseed

(milled flaxseed)

2013 research found that ground flaxseed had one of the strongest anti-hypertensive effects of any food. A few tablespoons (added to a smoothie or baked goods) lowered blood pressure and stroke risk 23 times more than daily aerobic exercise (6).

Flax shouldn't replace exercise, but its nutritional punch is worth adding to your diet.

5. Other seeds

(chia seeds, hemp seeds, pumpkin seeds, sesame seeds, fennel seeds)

Seeds are high in fiber and omega-3 fats and can be added to most dishes without being noticed.

When eaten with or after a meal, chia seeds moderate blood sugar and reduce inflammatory chemicals in the blood (7). Overall, a great daily addition.

6. Dates

Dates are one of the world's highest sugar foods, with 80% sugar by weight. Pure cake frosting is 60%, maple syrup is 66%, and cotton-candy jelly beans are 70%.

Despite their high sugar content, dates have a low glycemic index, meaning they don't affect blood sugar levels dramatically. They also improve triglyceride and antioxidant stress levels (8).

Dates are a great source of energy and contain high levels of dietary fiber and polyphenols, making 3-10 dates a great way to fight disease, support gut health with prebiotics, and satisfy a sweet tooth (9).

7. Cruciferous Veggies

(broccoli, Brussel sprouts, horseradish, kale, cauliflower, cabbage, boy choy, arugula, radishes, turnip greens)

Cruciferous vegetables contain an active ingredient that makes them disease-fighting powerhouses. Sulforaphane protects our brain, eyesight, against free radicals and environmental hazards, and treats and prevents cancer (10).

Unless you eat raw cruciferous vegetables daily, you won't get enough sulforaphane (and thus, its protective nutritional benefits). Cooking destroys the enzyme needed to create this super-compound.

If you chop broccoli, cauliflower, or turnip greens and let them sit for 45 minutes before cooking them, the enzyme will have had enough time to work its sulforaphane magic, allowing the vegetables to retain the same nutritional value as if eaten raw. Crazy, right? For more on this, see What Chopping Your Vegetables Has to Do with Fighting Cancer.

8. Whole grains

(barley, brown rice, quinoa, oats, millet, popcorn, whole-wheat pasta, wild rice)

Whole-grains are one of the healthiest ways to consume your daily carbs and help maintain healthy gut flora.

This happens when fibre is broken down in the colon and starts a chain reaction, releasing beneficial substances into the bloodstream and reducing the risk of Type 2 Diabetes and inflammation (11).

9. Spices

(turmeric, cumin, cinnamon, ginger, saffron, cloves, cardamom, chili powder, nutmeg, coriander)

7% of a person's cells will have DNA damage. This damage is caused by tiny breaks in our DNA caused by factors like free-radical exposure.

Free radicals cause mutations that damage lipids, proteins, and DNA, increasing the risk of disease and cancer. Free radicals are unavoidable because they result from cellular metabolism, but they can be avoided by consuming anti-oxidant and detoxifying foods.

Including spices and herbs like rosemary or ginger in our diet may cut DNA damage by 25%. Yes, this damage can be improved through diet. Turmeric worked better at a lower dose (just a pinch, daily). For maximum free-radical fighting (and anti-inflammatory) effectiveness, use 1.5 tablespoons of similar spices (12).

10. Leafy greens

(spinach, collard greens, lettuce, other salad greens, swiss chard)

Studies show that people who eat more leafy greens perform better on cognitive tests and slow brain aging by a year or two (13).

As we age, blood flow to the brain drops due to a decrease in nitric oxide, which prevents blood vessels from dilatation. Daily consumption of nitrate-rich vegetables like spinach and swiss chard may prevent dementia and Alzheimer's.

11. Fermented foods

(sauerkraut, tempeh, kombucha, plant-based kefir)

Miso, kimchi, and sauerkraut contain probiotics that support gut microbiome.

Probiotics balance the good and bad bacteria in our bodies and offer other benefits. Fermenting fruits and vegetables increases their antioxidant and vitamin content, preventing disease in multiple ways (14).

12. Sea vegetables

(seaweed, nori, dulse flakes)

A population study found that eating one sheet of nori seaweed per day may cut breast cancer risk by more than half (15).

Seaweed and sea vegetables may help moderate estrogen levels in the metabolism, reducing cancer and disease risk.

Sea vegetables make up 30% of the world's edible plants and contain unique phytonutrients. A teaspoon of these super sea-foods on your dinner will help fight disease from the inside out.

13. Water

I'm less concerned about whether you consider water food than whether you drink enough. If this list were ranked by what single item led to the best health outcomes, water would be first.

Research shows that people who drink 5 or more glasses of water per day have a 50% lower risk of dying from heart disease than those who drink 2 or less (16).

Drinking enough water boosts energy, improves skin, mental health, and digestion, and reduces the risk of various health issues, including obesity.

14. Tea

All tea consumption is linked to a lower risk of stroke, heart disease, and early death, with green tea leading for antioxidant content and immediate health benefits.

Green tea leaves may also be able to interfere with each stage of cancer formation, from the growth of the first mutated cell to the spread and progression of cancer in the body. Green tea is a quick and easy way to support your long-term and short-term health (17).

15. Supplemental B12 vitamin

B12, or cobalamin, is a vitamin responsible for cell metabolism. Not getting enough B12 can have serious consequences.

Historically, eating vegetables from untreated soil helped humans maintain their vitamin B12 levels. Due to modern sanitization, our farming soil lacks B12.

B12 is often cited as a problem only for vegetarians and vegans (as animals we eat are given B12 supplements before slaughter), but recent studies have found that plant-based eaters have lower B12 deficiency rates than any other diet (18).

Article Sources:

1. The Global Burden of Disease Study 2010 (GBD 2010)

2. I. Darmadi-Blackberry, M. Wahlqvist, A. Kouris-Blazos, et al. Legumes: the most important dietary predictor of survival in older people of different ethnicities. Asia Pac J Clin Nutr. 2004;13(2):217–20.

3. Guha N, Kwan ML, Quesenberry CP Jr, Weltzien EK, Castillo AL, Caan BJ. Soy isoflavones and risk of cancer recurrence in a cohort of breast cancer survivors: the Life After Cancer Epidemiology study. Breast Cancer Res Treat. 2009 Nov;118(2):395–405.

4. Y. Bao, J. Han, F. B. Hu, E. L. Giovannucci, M. J. Stampfer, W. C. Willett, C. S. Fuchs. Association of nut consumption with total and cause-specific mortality. N. Engl. J. Med. 2013 369(21):2001–2011.

5. V. Vadivel, C. N. Kunyanga, H. K. Biesalski. Health benefits of nut consumption with special reference to body weight control. Nutrition 2012 28(11–12):1089–1097.

6. D Rodriguez-Leyva, W Weighell, A L Edel,R LaVallee, E Dibrov,R Pinneker, T G Maddaford, B Ramjiawan, M Aliani, R Guzman R, G N Pierce. Potent antihypertensive action of dietary flaxseed in hypertensive patients. Hypertension. 2013 Dec;62(6):1081–9. doi: 10.1161/HYPERTENSIONAHA.113.02094.

7. Vuksan V, Jenkins AL, Dias AG, Lee AS, Jovanovski E, Rogovik AL, Hanna A. Reduction in postprandial glucose excursion and prolongation of satiety: possible explanation of the long-term effects of whole grain Salba (Salvia Hispanica L.). Eur J Clin Nutr. 2010 Apr;64(4):436–8. doi: 10.1038/ejcn.2009.159. Epub 2010 Jan 20. PMID: 20087375.

8. W. Rock, M. Rosenblat, H. Borochov-Neori, N. Volkova, S. Judeinstein, M. Elias, and M. Aviram. Effects of date (Phoenix dactylifera L., Medjool or Hallawi Variety) consumption by healthy subjects on serum glucose and lipid levels and on serum oxidative status: a pilot study. J. Agric. Food. Chem., 57(17):8010{8017, 2009.

9. Eid N, Enani S, Walton G, et al. The impact of date palm fruits and their component polyphenols, on gut microbial ecology, bacterial metabolites and colon cancer cell proliferation. J Nutr Sci. 2014;3:e46.

10. Li Y, Zhang T, Korkaya H, Liu S, Lee HF, Newman B, Yu Y, Clouthier SG, Schwartz SJ, Wicha MS, Sun D. Sulforaphane, a Dietary Component of Broccoli/Broccoli Sprouts, Inhibits Breast Cancer Stem Cells. Clin Cancer Res. 2010 May 1;16(9):2580–90.

11. Lappi J, Kolehmainen M, Mykkänen H, Poutanen K. Do large intestinal events explain the protective effects of whole grain foods against type 2 diabetes? Crit Rev Food Sci Nutr. 2013;53(6):631–40.

12. S. S. Percival, J. P. V. Heuvel, C. J. Nieves, C. Montero, A. J. Migliaccio, J. Meadors. Bioavailability of Herbs and Spices in Humans as Determined by ex vivo Inflammatory Suppression and DNA Strand Breaks. J Am Coll Nutr. 2012 31(4):288–294.

13. Nurk E, Refsum H, Drevon CA, et al. Cognitive performance among the elderly in relation to the intake of plant foods. The Hordaland Health Study. Br J Nutr. 2010;104(8):1190–201.

14. Melini, F.; Melini, V.; Luziatelli, F.; Ficca, A.G.; Ruzzi, M. Health-Promoting Components in Fermented Foods: An Up-to-Date Systematic Review. Nutrients2019, 11, 1189.

15. H. Funahashi, T. Imai, T. Mase, M. Sekiya, K. Yokoi, H. Hayashi, A. Shibata, T. Hayashi, M. Nishikawa, N. Suda, Y. Hibi, Y. Mizuno, K. Tsukamura, A. Hayakawa, S. Tanuma. Seaweed prevents breast cancer? Jpn. J. Cancer Res. 2001 92(5):483–487.

16. Chan J, Knutsen SF, Blix GG, Lee JW, Fraser GE. Water, other fluids, and fatal coronary heart disease: the Adventist Health Study. Am J Epidemiol. 2002 May 1;155(9):827–33. doi: 10.1093/aje/155.9.827. PMID: 11978586.

17. Fujiki H, Imai K, Nakachi K, Shimizu M, Moriwaki H, Suganuma M. Challenging the effectiveness of green tea in primary and tertiary cancer prevention. J Cancer Res Clin Oncol. 2012 Aug;138(8):1259–70.

18. Damayanti, D., Jaceldo-Siegl, K., Beeson, W. L., Fraser, G., Oda, K., & Haddad, E. H. (2018). Foods and Supplements Associated with Vitamin B12Biomarkers among Vegetarian and Non-Vegetarian Participants of the Adventist Health Study-2 (AHS-2) Calibration Study. Nutrients, 10(6), 722. doi:10.3390/nu10060722

My wallpape

Deep learning-based generative art is being researched. As usual, self-learning is better. Some models, like OpenAI's DALL-E 2, require registration and can only be used online, but others can be used locally, which is usually more enjoyable for curious users. I'll demonstrate the Stable Diffusion model's operation on a standard PC.

Let’s get started.

What It Does

Stable Diffusion uses numerous components:

• A generative model trained to produce images is called a diffusion model. The model is incrementally improving the starting data, which is only random noise. The model has an image, and while it is being trained, the reversed process is being used to add noise to the image. Being able to reverse this procedure and create images from noise is where the true magic is (more details and samples can be found in the paper).

• An internal compressed representation of a latent diffusion model, which may be altered to produce the desired images, is used (more details can be found in the paper). The capacity to fine-tune the generation process is essential because producing pictures at random is not very attractive (as we can see, for instance, in Generative Adversarial Networks).

• A neural network model called CLIP (Contrastive Language-Image Pre-training) is used to translate natural language prompts into vector representations. This model, which was trained on 400,000,000 image-text pairs, enables the transformation of a text prompt into a latent space for the diffusion model in the scenario of stable diffusion (more details in that paper).

This figure shows all data flow:

The weights file size for Stable Diffusion model v1 is 4 GB and v2 is 5 GB, making the model quite huge. The v1 model was trained on 256x256 and 512x512 LAION-5B pictures on a 4,000 GPU cluster using over 150.000 NVIDIA A100 GPU hours. The open-source pre-trained model is helpful for us. And we will.

Install

Before utilizing the Python sources for Stable Diffusion v1 on GitHub, we must install Miniconda (assuming Git and Python are already installed):

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh
chmod +x Miniconda3-py39_4.12.0-Linux-x86_64.sh
./Miniconda3-py39_4.12.0-Linux-x86_64.sh
conda update -n base -c defaults conda

Install the source and prepare the environment:

git clone https://github.com/CompVis/stable-diffusion
cd stable-diffusion
conda env create -f environment.yaml
conda activate ldm
pip3 install transformers --upgrade

Download the pre-trained model weights next. HiggingFace has the newest checkpoint sd-v14.ckpt (a download is free but registration is required). Put the file in the project folder and have fun:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Almost. The installation is complete for happy users of current GPUs with 12 GB or more VRAM. RuntimeError: CUDA out of memory will occur otherwise. Two solutions exist.

Running the optimized version

Try optimizing first. After cloning the repository and enabling the environment (as previously), we can run the command:

python3 optimizedSD/optimized_txt2img.py --prompt "hello world" --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Stable Diffusion worked on my visual card with 8 GB RAM (alas, I did not behave well enough to get NVIDIA A100 for Christmas, so 8 GB GPU is the maximum I have;).

Running Stable Diffusion without GPU

If the GPU does not have enough RAM or is not CUDA-compatible, running the code on a CPU will be 20x slower but better than nothing. This unauthorized CPU-only branch from GitHub is easiest to obtain. We may easily edit the source code to use the latest version. It's strange that a pull request for that was made six months ago and still hasn't been approved, as the changes are simple. Readers can finish in 5 minutes:

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available at line 20 of ldm/models/diffusion/ddim.py ().

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available in line 20 of ldm/models/diffusion/plms.py ().

• Replace device=cuda in lines 38, 55, 83, and 142 of ldm/modules/encoders/modules.py with device=cuda if torch.cuda.is available(), otherwise cpu.

• Replace model.cuda() in scripts/txt2img.py line 28 and scripts/img2img.py line 43 with if torch.cuda.is available(): model.cuda ().

Run the script again.

Testing

Test the model. Text-to-image is the first choice. Test the command line example again:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

The slow generation takes 10 seconds on a GPU and 10 minutes on a CPU. Final image:

Hello world is dull and abstract. Try a brush-wielding hamster. Why? Because we can, and it's not as insane as Napoleon's cat. Another image:

Generating an image from a text prompt and another image is interesting. I made this picture in two minutes using the image editor (sorry, drawing wasn't my strong suit):

I can create an image from this drawing:

python3 scripts/img2img.py --prompt "A bird is sitting on a tree branch" --ckpt sd-v1-4.ckpt --init-img bird.png --strength 0.8

It was far better than my initial drawing:

I hope readers understand and experiment.

Stable Diffusion UI

Developers love the command line, but regular users may struggle. Stable Diffusion UI projects simplify image generation and installation. Simple usage:

• Unpack the ZIP after downloading it from https://github.com/cmdr2/stable-diffusion-ui/releases. Linux and Windows are compatible with Stable Diffusion UI (sorry for Mac users, but those machines are not well-suitable for heavy machine learning tasks anyway;).

• Start the script.

Done. The web browser UI makes configuring various Stable Diffusion features (upscaling, filtering, etc.) easy:

V2.1 of Stable Diffusion

I noticed the notification about releasing version 2.1 while writing this essay, and it was intriguing to test it. First, compare version 2 to version 1:

• alternative text encoding. The Contrastive LanguageImage Pre-training (CLIP) deep learning model, which was trained on a significant number of text-image pairs, is used in Stable Diffusion 1. The open-source CLIP implementation used in Stable Diffusion 2 is called OpenCLIP. It is difficult to determine whether there have been any technical advancements or if legal concerns were the main focus. However, because the training datasets for the two text encoders were different, the output results from V1 and V2 will differ for the identical text prompts.

• a new depth model that may be used to the output of image-to-image generation.

• a revolutionary upscaling technique that can quadruple the resolution of an image.

• Generally higher resolution Stable Diffusion 2 has the ability to produce both 512x512 and 768x768 pictures.

The Hugging Face website offers a free online demo of Stable Diffusion 2.1 for code testing. The process is the same as for version 1.4. Download a fresh version and activate the environment:

conda deactivate  
conda env remove -n ldm  # Use this if version 1 was previously installed
git clone https://github.com/Stability-AI/stablediffusion
cd stablediffusion
conda env create -f environment.yaml
conda activate ldm

Hugging Face offers a new weights ckpt file.

The Out of memory error prevented me from running this version on my 8 GB GPU. Version 2.1 fails on CPUs with the slow conv2d cpu not implemented for Half error (according to this GitHub issue, the CPU support for this algorithm and data type will not be added). The model can be modified from half to full precision (float16 instead of float32), however it doesn't make sense since v1 runs up to 10 minutes on the CPU and v2.1 should be much slower. The online demo results are visible. The same hamster painting with a brush prompt yielded this result:

It looks different from v1, but it functions and has a higher resolution.

The superresolution.py script can run the 4x Stable Diffusion upscaler locally (the x4-upscaler-ema.ckpt weights file should be in the same folder):

python3 scripts/gradio/superresolution.py configs/stable-diffusion/x4-upscaling.yaml x4-upscaler-ema.ckpt

This code allows the web browser UI to select the image to upscale:

The copy-paste strategy may explain why the upscaler needs a text prompt (and the Hugging Face code snippet does not have any text input as well). I got a GPU out of memory error again, although CUDA can be disabled like v1. However, processing an image for more than two hours is unlikely:

Stable Diffusion Limitations

When we use the model, it's fun to see what it can and can't do. Generative models produce abstract visuals but not photorealistic ones. This fundamentally limits The generative neural network was trained on text and image pairs, but humans have a lot of background knowledge about the world. The neural network model knows nothing. If someone asks me to draw a Chinese text, I can draw something that looks like Chinese but is actually gibberish because I never learnt it. Generative AI does too! Humans can learn new languages, but the Stable Diffusion AI model includes only language and image decoder brain components. For instance, the Stable Diffusion model will pull NO WAR banner-bearers like this:

V1:

V2.1:

The shot shows text, although the model never learned to read or write. The model's string tokenizer automatically converts letters to lowercase before generating the image, so typing NO WAR banner or no war banner is the same.

I can also ask the model to draw a gorgeous woman:

V1:

V2.1:

The first image is gorgeous but physically incorrect. A second one is better, although it has an Uncanny valley feel. BTW, v2 has a lifehack to add a negative prompt and define what we don't want on the image. Readers might try adding horrible anatomy to the gorgeous woman request.

If we ask for a cartoon attractive woman, the results are nice, but accuracy doesn't matter:

V1:

V2.1:

Another example: I ordered a model to sketch a mouse, which looks beautiful but has too many legs, ears, and fingers:

V1:

V2.1: improved but not perfect.

V1 produces a fun cartoon flying mouse if I want something more abstract:

I tried multiple times with V2.1 but only received this:

The image is OK, but the first version is closer to the request.

Stable Diffusion struggles to draw letters, fingers, etc. However, abstract images yield interesting outcomes. A rural landscape with a modern metropolis in the background turned out well:

V1:

V2.1:

Generative models help make paintings too (at least, abstract ones). I searched Google Image Search for modern art painting to see works by real artists, and this was the first image:

I typed "abstract oil painting of people dancing" and got this:

V1:

V2.1:

It's a different style, but I don't think the AI-generated graphics are worse than the human-drawn ones.

The AI model cannot think like humans. It thinks nothing. A stable diffusion model is a billion-parameter matrix trained on millions of text-image pairs. I input "robot is creating a picture with a pen" to create an image for this post. Humans understand requests immediately. I tried Stable Diffusion multiple times and got this:

This great artwork has a pen, robot, and sketch, however it was not asked. Maybe it was because the tokenizer deleted is and a words from a statement, but I tried other requests such robot painting picture with pen without success. It's harder to prompt a model than a person.

I hope Stable Diffusion's general effects are evident. Despite its limitations, it can produce beautiful photographs in some settings. Readers who want to use Stable Diffusion results should be warned. Source code examination demonstrates that Stable Diffusion images feature a concealed watermark (text StableDiffusionV1 and SDV2) encoded using the invisible-watermark Python package. It's not a secret, because the official Stable Diffusion repository's test watermark.py file contains a decoding snippet. The put watermark line in the txt2img.py source code can be removed if desired. I didn't discover this watermark on photographs made by the online Hugging Face demo. Maybe I did something incorrectly (but maybe they are just not using the txt2img script on their backend at all).

Conclusion

The Stable Diffusion model was fascinating. As I mentioned before, trying something yourself is always better than taking someone else's word, so I encourage readers to do the same (including this article as well;).

Is Generative AI a game-changer? My humble experience tells me:

• I think that place has a lot of potential. For designers and artists, generative AI can be a truly useful and innovative tool. Unfortunately, it can also pose a threat to some of them since if users can enter a text field to obtain a picture or a website logo in a matter of clicks, why would they pay more to a different party? Is it possible right now? unquestionably not yet. Images still have a very poor quality and are erroneous in minute details. And after viewing the image of the stunning woman above, models and fashion photographers may also unwind because it is highly unlikely that AI will replace them in the upcoming years.

• Today, generative AI is still in its infancy. Even 768x768 images are considered to be of a high resolution when using neural networks, which are computationally highly expensive. There isn't an AI model that can generate high-resolution photographs natively without upscaling or other methods, at least not as of the time this article was written, but it will happen eventually.

• It is still a challenge to accurately represent knowledge in neural networks (information like how many legs a cat has or the year Napoleon was born). Consequently, AI models struggle to create photorealistic photos, at least where little details are important (on the other side, when I searched Google for modern art paintings, the results are often even worse;).

• When compared to the carefully chosen images from official web pages or YouTube reviews, the average output quality of a Stable Diffusion generation process is actually less attractive because to its high degree of randomness. When using the same technique on their own, consumers will theoretically only view those images as 1% of the results.

Anyway, it's exciting to witness this area's advancement, especially because the project is open source. Google's Imagen and DALL-E 2 can also produce remarkable findings. It will be interesting to see how they progress.