[PhilBa]

The recent thread on the Roman Potter's tray got me thinking about how to take a photo (or painting) and come up with a design of the pictured object. I do this sort of thing fairly often and a number of comments made me realize there is interest in how one might do that. The following post is an abbreviated version of my blog post here.

If you've been woodworking for a while, you've probably had someone come to you with a picture asking "can you make this?" Or perhaps your spouse shows you a picture of a piece of furniture that is wildly expensive but doesn't look that hard to make. If nothing else, it's an opportunity to acquire a tool with spousal approval! But how do you go from picture to plans?

First, you need to have some CAD skills. You don't need to be a maestro at it but do need to be able make a simple drawing and add dimensions. Here are the basic steps:

1. Modify the photo to correct for perspective distortion.
2. Find a reference object of known (or knowable) size. In the Roman Potter thread, I used the guy's arm as a reference.
3. Scale the corrected photo so the reference object is the correct size.
4. Dimension the important features.
5. "Normalize" the dimension so they are consistent with standard sizes.
6. Trace out any profiles for later use in modeling.
7. Use the dimensions and profile(s) to model the object.

In the photos, you can see a picture of an end-table I own. I applied perspective correction to the photo (see blog for more details on what software you could use). Imported the photo into CAD software. Usually you will have overall dimensions so you can use that to scale. In this case the table is 26x17x25. Dimensioned the width and height and then scaled so the photo is the proper size. Then added dimensions of the various features. Then adjusted (normalized) the dimensions so they are consistent with standard lumber dimensions and are consistent with each other. Like rails are likely all the same size. Also, trace any profiles that you will need to cut.

From there, you should be able to model the object with a good degree of fidelity.

 

[difalkner]

All good info, Phil. Thanks!

David

 

[JOAT]

First, you need to have some CAD skills.

Not actually. I get a lot of inspiration from pictures, for a lot of my projects. Including some furniture pieces I will eventually get around to. All I use are a pencil, eraser, graph paper, and possibly a ruler. I figure out the dimensions after I have something laid out that I like the looks of. I've been making stuff with wood long before CAD was even thought of, didn't need it then, don't need it now. I actually sketch a rough picture of a lot of things, in spiral notebooks, but nothing like 1/4" graph paper for laying out a finished design.

 

[PhilBa]

Not actually. I get a lot of inspiration from pictures, for a lot of my projects. Including some furniture pieces I will eventually get around to. All I use are a pencil, eraser, graph paper, and possibly a ruler. I figure out the dimensions after I have something laid out that I like the looks of. I've been making stuff with wood long before CAD was even thought of, didn't need it then, don't need it now. I actually sketch a rough picture of a lot of things, in spiral notebooks, but nothing like 1/4" graph paper for laying out a finished design.

For my method, you do. I'd like to see a post describing your's in detail, though.

 

[sreilly]

Interesting thread Phil using a well thought out process. Out of curiosity what did you use for the scaling. I was thinking the coasters are usually a fairly standard size. I guess these tables also have a fairly standard height? I have and am fairly fluent in AutoCAD (older version) but not familiar with importing pictures. I'll have to look at that. Mostly I used it for mechanical HVAC system ductwork design and custom control schematics.I also have a tendency to draw out my projects before hand. Seems to let me make less mistakes when I know ahead what is being done. A lot like being woken up at 3am by the squad tones (in building dorm with lights coming on and announcement over the speaker) and hearing this distorted voice giving the address and details of the call. Makes no sense until I write down the address and then I remember it clearly seldom transposing the address numbers.

I'll read the blog and maybe save it as a PDF for future reference if that's permissible.

 

[PhilBa]

Interesting thread Phil using a well thought out process. Out of curiosity what did you use for the scaling. I was thinking the coasters are usually a fairly standard size. I guess these tables also have a fairly standard height? I have and am fairly fluent in AutoCAD (older version) but not familiar with importing pictures. I'll have to look at that. Mostly I used it for mechanical HVAC system ductwork design and custom control schematics.I also have a tendency to draw out my projects before hand. Seems to let me make less mistakes when I know ahead what is being done. A lot like being woken up at 3am by the squad tones (in building dorm with lights coming on and announcement over the speaker) and hearing this distorted voice giving the address and details of the call. Makes no sense until I write down the address and then I remember it clearly seldom transposing the address numbers.

I'll read the blog and maybe save it as a PDF for future reference if that's permissible.

If you are reproducing a commercial product from an ad, they almost always have overall measurements. Also, museums often list measurements. But lacking that, you have the right idea. Also, there are standard heights, widths, lengths that can be used. For example, kitchen counters are almost always 36"/90cm high give or take a few percent. Tables and desks are typically 30"/76cm high. There are standards on the internet for lots of things. Here's one for furniture. One for human males.

One thing that takes a little guessing is if your "yardstick" isn't in the same viewing plane as the object front (or side if that's what you are modeling) then you have adjust it's actual size. It's intuitive, though - farther away is smaller than it should be, closer is larger.

On drawing designs in CAD. I don't always do that but, for anything requiring joinery, it makes a huge difference. I can see how the joints go together and look at the object from multiple points of view. Joinery can benefit a lot - there was a project I did where I was able to see two tenons interfering with each other and repositioned them so they didn't. Better to run into that kind of thing on the computer than in the shop after making some cuts. You also get shop drawings easily. Another benefit is you can create multiple versions to play with look and fit. And, as I've said multiple times, get buy-in from the "customer". You can also create cut lists/plans which is particularly great for working with sheet material. On my big 15 drawer shop cabinet project, I was able to be really efficient in the use of plywood - only had a small amount of leftover wood. On my Plyometric boxes, I discovered that if I made them 1/4" smaller I avoided having to buy an extra sheet of Baltic Birch which is really expensive - saved about $80 on that though the excess wouldn't have gone to waste. No way I could have done that with paper and a sketch book. Perhaps there is a good discussion to be had for this topic.

 

[woodworker47]

Phil,

Thanks for posting this. I often use AutoCad to make drawings of the project I am building. I then make a cut list sorted by material thickness. Since I resaw my Walnut to make boards, this helps me keep track of board inventory for a project.

I will also look into your blog to learn more about bringing photos into a Cad software. I have a Cousin that has done this, but he lives to far away to teach me the process. Several times I have viewed a picture of an item I wanted to build, but had a hard time with dimensions. Hopefully your blog is what I needed to learn process.

Frank

 

[JOAT]

For my method, you do. I'd like to see a post describing your's in detail, though.

Actually, I use several methods. None involving CAD or CNC machines (my CNC is Coffee n Cookies - yesterday was coffe and some very tasty chochalet chip cookies). I'll break down the different methods, that will keep me less confused.

The first method is Zen Woodworking. I need something, no idea how to make whatever, so head for the shop, sit down, grab a piece of wood, and the next thing I actually realize is I've made whatever. And it works. No picture on this one, but did my saw stand that way. Turned out sort of art deco, and I think I could do one again. The first one was a bit tall, so used a precision tool and cut it a bit shorter (chainsaw), glued it back together, and just right. The pictures are of pieces I made the same way. Except for the tool stand, made that one from small pieces of leftover wood, over a period of several days. The router table is something over 15 years old, and I still don't know how I did it. The top is still completely flat, the router plate is also 1/2" plywood and fits in perfectly. Not a clue where I got the idea. Did the same with the small metal bandsaw stand, but this one I am pretty sure I could duplicate again, if I had to. Then there are hooks for hanging power cords on. Needed a chair, so made one. Another easy to duplicate. Later cut the back off the chair, and made a small workbench out of it - too hard to be comfortable sitting for long, and too much bother moving it, and needed a bench.

No plans, no sketches, for any of these.

 

[JOAT]

Drawing method

OK, these were all hand drawn. Inspired by a lot of pictures, a LOT of pictures.

The Wizard was inspired by Gandolf, Merlin, Getafix the Druid wizard, the wizards of The Unseen University, and some imagination. The original had no brim on his hat, and later I added one. The original design was rough sketched, over and over, until I got what I liked, then fine tuned on 1/4" graph paper. This was cut out, glued to plywood, cut and sanded to a final size, then a master made with it.

The cigar store Indian was done about the same. The face was originally inspired by a photo from about 1911 of a member of one of the North West tribes. Then about 11 changes made to the fact. Each part of the clothing and war bonnet came from photographs of actual historical pieces. Sketch, erase, sketch, erase. This one is about 2' tall, and am working on one that will be around 5' tall. Didn't use graph paper on this one, just laid it out by hand once I got a design I liked. That particular piece was given to my older son. Then found out he had stuck it in a shed. Made him get it out, and now it is on his porch beside his front door. But still can't get him to clean it up.

The International Company Logo eagle, that is over 20 years now, so not 100% sure just how that was made, but thinking also done by hand, from a photo of an antique off of a very old sailing ship.

The fairy was also hand drawn, then painted, for my granddaughter. This was a PITA, and unlikely anymore will be done.

 

[PhilBa]

The kind of stuff you do doesn't really need CAD. Anything with a lot of parts or moving parts or complex joinery benefits significantly, though.

 

[JOAT]

The kind of stuff you do doesn't really need CAD. Anything with a lot of parts or moving parts or complex joinery benefits significantly, though.

Yep, but I'm not done yet.

 

[JOAT]

Monster truck bank

This is a bit different. I don't recall for sure (it's been awhile) but think their are about 6 different shapes in this. It's my first monster truck bank design, made for my older grandson. The drawing is a rough sketch, came up with saving and viewing a lot of monster trucks. The final was laid out on 1/4" graph paper, full size. There were no individual parts drawings. I just glue down the final drawing, cut out the parts one piece at a time from that. Those pieces are the first step in making my masters. Those are sanded to final, glued to a piece of wood, routed, and are then masters and I can duplicate exact pieces from them. My trucks designs are more detailed now, as well as simpler. My piggy banks are done similarly. My method confuses a lot of people, but I find it pretty simple. Oh yes, I decide the length and height of these, mark those, then it's basically freehand the details, using a straightedge where appropriate. The "plans" are done when I'm happy with them - which is why a good eraser is a necessity.

 

[JOAT]

Owl house signs

Once I get my son to weld up a holder for me (I will likely have to do it myself if I want it done this century), I will be making house signs, with the house numbers, and placing them at the end of my drive. There will be various designs, and they will get changed every once in awhile. Decided to make an owl or 2 for one designs. The house numbers will probably be below the figures. When I wound up with the design I liked, discarded all the owl photos I had gone thru - well over 300 owl photos - I tend to do a lot of research on a lot of my projects.

Here is the final design, and a couple of friendly owls that will go on the signs. The signs will likely be around 2 foot tall, not including the house numbers - subject to change.

Oh yeah, the friendly Wizard will also be one of the house signs. Still researching figures for house signs. All friendly of course.
>

 

[JOAT]

Raindeer yard art

Around 20 years or so ago, my then daughter-in-law pestered me to make her 2 of these raindeer. Finally said yes, to keep her quiet. At the time the plans I found were $20, each, for a total of $40. And no way was she going to pay. So I found a photo online of the parts all laid out. Printed that out, then laid out a 1/4" graph in pencil on the page. The deer were supposed to be almost 4' tall, so figured out that a 4" graph on plywood would be perfect. Just looked at the printed page, drew out the plans on the wood by copying what was in a 1/4" square into a 4" square. Came out perfect, and didn't take as long as I had expected. Cut them out, put them together, perfect. No mention was made, but best to drill a hole thru the legs, so you can drive a tent peg into the ground and secure the deer to those, otherwise they blow over very easily.

Had thought of making more, to sell, but found out after just that one pair that I absolutely hated making them, so didn't make any. Might have been because of my ex dau-in-law, don't know.

Very simple, and effective, for enlarging plans. Oh yeah, I had to paint the damn things too.

 

[JOAT]

3 cup drink carrier

Most Saturdays I go visit my older son and his two. And wind up buying lunch from one of the fast food joints. Got very tired of their floppy drink carriers, so made one. Looked at a few pictures, no way did I feel like making one out of wood, so took out my pocket knife, got cardboard, Titebond II, cloth to cover it, and started in. Not very pretty, but very functional, and the people at the food joint drive thru windows love it. Forgot it at my son's one day, so decided I'm make a backup. The first one took several mods before it was right, so drew plans of it, and commenced with #2. Decided to decorate that one a bit, so glued cards on all sides. Used Fruit of The Loom boxer waistbands as handles - which really set my son off. Hilarious. So I covered the handles with cloth. I have even had several offers to buy, and may make a few more, just to sell. I don't normally make things out of cardboard, so just thought I'd post this for laughs.

This is not the order the were posted in.

 

[JOAT]

Chair from a picture

I think I'll just call this off after this is posted.

I saw how you did the dimensions of the table. I would do it different. I would figure out how high, how wide, how deep I wanted it, and go from there. The picture would just be for inspiration. But don't do a lot of furniture, let alone copy someone else's work, so not making anything similar.

However, if I were to make a copy of this chair, it is unlikely I would try to copy it exactly, but if I did, I figure it would be simple enough. Basically just blow up the side piece, by making a graph, make copies. The however is: However, it is quite likely I would not care for the curves in the seat, in fact I can almost guarantee it. So, grab some cardboard, sketch out curves, repeat until I get something I figure would be comfortable to me. Then I'd sketch out the outline of a whole piece, glue that to wood, cut it, and make a master. Then make sufficient copies to make the chair pieces. Do the same with the two supports. Glue it all together, and try it. Then if it is not quite what I wanted, I'd sell it, and make another. Simple.

 

[PhilBa]

You can always find exceptions to any rule.

The reason why you would want to copy the curves of the chair is that they are probably based on some amount of experience and/or research into what makes a comfortable outline. To get it right, you would need to use some sort of perspective correction - just blowing it up and tracing doesn't get you there. Though, in the case of that picture, there isn't anything visible to use as a reference. Overall dimensions could be used to get pretty close. No question, this is a hard one to use as a reference design.

 

[JOAT]

You can always find exceptions to any rule.

The reason why you would want to copy the curves of the chair is that they are probably based on some amount of experience and/or research into what makes a comfortable outline. To get it right, you would need to use some sort of perspective correction - just blowing it up and tracing doesn't get you there. Though, in the case of that picture, there isn't anything visible to use as a reference. Overall dimensions could be used to get pretty close. No question, this is a hard one to use as a reference design.

Nah, I actually would not want to copy the curves of that chair. First because I don't want to make a copy of the chair. Second, way I figure it, the chair is contoured to the designer, not me. Blowing it up, and tracing, is only a start, so I could start on designing curves that would be comfortable for me, THAT is what would get me there. Certainly there is something visible to use as a reference, the picture, and the idea of the curves. I'd just design it the length I would want, and width I would want, I don't much care about measurements otherwise. If I made something like that (don't plan on it) the curves would be custom fit to me. I chose this chair as an example because I think it would be simple - the only part not so simple would be getting the curves fit to me, likely take some time on that, after that, easy peasy.

 

[Laserwizard]

The recent thread on the Roman Potter's tray got me thinking about how to take a photo (or painting) and come up with a design of the pictured object. I do this sort of thing fairly often and a number of comments made me realize there is interest in how one might do that. The following post is an abbreviated version of my blog post here.

If you've been woodworking for a while, you've probably had someone come to you with a picture asking "can you make this?" Or perhaps your spouse shows you a picture of a piece of furniture that is wildly expensive but doesn't look that hard to make. If nothing else, it's an opportunity to acquire a tool with spousal approval! But how do you go from picture to plans?

First, you need to have some CAD skills. You don't need to be a maestro at it but do need to be able make a simple drawing and add dimensions. Here are the basic steps:

1. Modify the photo to correct for perspective distortion.
2. Find a reference object of known (or knowable) size. In the Roman Potter thread, I used the guy's arm as a reference.
3. Scale the corrected photo so the reference object is the correct size.
4. Dimension the important features.
5. "Normalize" the dimension so they are consistent with standard sizes.
6. Trace out any profiles for later use in modeling.
7. Use the dimensions and profile(s) to model the object.

In the photos, you can see a picture of an end-table I own. I applied perspective correction to the photo (see blog for more details on what software you could use). Imported the photo into CAD software. Usually you will have overall dimensions so you can use that to scale. In this case the table is 26x17x25. Dimensioned the width and height and then scaled so the photo is the proper size. Then added dimensions of the various features. Then adjusted (normalized) the dimensions so they are consistent with standard lumber dimensions and are consistent with each other. Like rails are likely all the same size. Also, trace any profiles that you will need to cut.

From there, you should be able to model the object with a good degree of fidelity.

I built this table by scaling it off of this picture. Came out really good.

 

[sreilly]

You indeed got excellent results.