Now the bamboo itself contains a range of colors from light to medium to dark brown, and mimicking the exact look would be nearly impossible. This is why its always important to test your stains on an actual scrap piece of the material you plan to work with.
But all of them were, of course, affected by the red oak laying underneath the color. The differences were subtle: some had more red, some had more brown, others had more yellow.
From left to right we have oil-based Pecan, oil-based Antique Cherry, American Cherry gel stain, Nutmeg gel stain, Light Brown water-based dye stain, and Pecan water-based wood stain. We tested numerous water and oil based stains and dyes (all General Finishes). But before diving into that craziness, we decided to take some dyes and stains from the cabinet and see if we couldn’t find a pre-made solution. So we were all prepared to pull out the pigments, dyes, lacquer and the HVLP. So if you are ever doing a match for a client, its incredibly important to manage their expectations. I would match the color perfectly, only to find the customer disappointed on delivery day because the pieces don’t look the same. I have had more than one customer ask me to refinish something made from pine so that it looks like something else made from oak. This is an important detail that many clients don’t realize. Now lets get one thing straight: there isn’t a damn thing you can do to oak to make it look like bamboo. Pretty reasonable, right? Well anyway, he’s building a custom piece for a client who wants the color to match a particular bamboo cutting board. Well, to tell you the truth, that’s my standard color matching consultation fee. This weekend, my buddy Ron from RJones Woodworks stopped by to bring me Dunkin’ Donuts coffee. The formula should always (hopefully) be consistent and if you ever need to reproduce the color again in the future, you’ll thank yourself. And if you can find the perfect color in a commercial product, I say “why not?”. Over the years, I began to realize that many times the perfect match is sitting in a can on the shelf. We used to mix all kind of crazy stuff together to get that perfect match. I was fortunate to work in a refinishing shop for a while and I had the opportunity to learn about color matching using various techniques and materials. But does it always have to be this convoluted? Let me spoil the ending for you: NO! There are so many different ways to arrive at a final color and look that it can drive you nuts! Lets see, there are alcohol and water soluble dyes in liquid and powder form, oil stains, water-based stains, pigments, toners, gel stains, glazes, and the list goes on and on. Here are examples of what the model considers "red," "green," "blue," and "purple.In my opinion, color matching is something of an art form. The model that learned to map names to colors is a straightforward convolutional neural network, and it seems to work pretty well. Using these word embeddings, I built two different models: one that maps a name to a color, and one that maps a color to a name. For more information on word embeddings and why they’re useful, check out my introduction to representation learning. (I used these word2vec embeddings.) The advantage of using pre-trained word embeddings is that the model doesn't have to learn as much about language during training it just needs to learn how different nouns and colors map to the RGB space and how adjectives should modify colors (e.g., "Analytical Gray"). With that in mind, I decided to build a model that mapped colors to names (and vice-versa) using pre-trained word embeddings. (This is an example of why open science is important when doing non-trivial things!) My suspicion is that the model was mostly trying to learn a character-level language model from the data, so regardless of whether there are 1,500 named colors or 7,700, that's definitely not enough words and phrases to learn anything meaningful about the English language. The first thing I noticed about the color data was that there were only about 1,500 named colors, so there's a discrepancy between the data Janelle was using and what I found.
I wondered whether the idea could be improved, so, exclusively using open source software, I built a model to play around with the code for the project can be found on my GitHub. A lot of the color names were nonsensical (i.e., not actual words), and the pairings between the generated names and the colors seemed pretty random, so it wasn't clear whether the model was actually learning a meaningful function to map colors to language. I thought the idea was really neat, but I found the results underwhelming. On a recent trip to Michigan, my friend Tim Sosa mentioned a blog post he read in which the author, Janelle Shane, describes how she built a neural network that generates color names from RGB values.