Wikipedia talk:WikiProject Chemistry/Structure drawing workgroup/Archive Apr 2007

Just to kick things off, I'll post a few comments.

1. I have tried the ChemDraw settings suggested in the style guide and found that the images are huge. The settings can be used for drawing a single structure of a small molecule, but even there I find the font ratio poor in many cases.
2. I think for general article info we may want to emulate books (such as Wade, Carey, etc) rather than journal settings. In other words, larger images, use of colour to explain points. For more advanced topics and structures involving more complex molecules, we should use one of the journal settings such as ACS or RSC. For a previous discussion along these lines see this and the related article at Aldol_reaction. (This article highlights another issue, how to include pictures and drawings side by side!)
3. It is regarded as good practice to create pictures larger than they are typically used in our articles. However, this creates problems. How can we make all settings on the screen the same (say, ACS) style, if this one is scaled differently from that one? Also, for small settings like ACS the atom labels are large in comparison to the bond length for clarity, whereas for larger book-style settings the atom labels are comparitively smaller - so how will these things scale if people use the same image at different sizes?
4. Whatever settings we choose, we should make it easy for new people to put those settings into their molecule editor, whatever they use. Many contributors don't have ChemDraw and are restricted to a variety of freeware.

Comments, anyone? Walkerma 06:31, 9 April 2006 (UTC)[reply]

Chemdraw and Chemsketch (and probably other programs too) allow to save drawing settings as a template. Ideally, the program used to draw structures should be irrelevant as long as we come up with a set of templates for each one and make them easily available to editors. -- Rune Welsh | ταλκ 14:10, 9 April 2006 (UTC)[reply]

The ISIS version is free for private use, but I do not know if Wikipedia is included with this licence. If yes, a free program would be a good thing - then everybody can download and draw images. --Stone 10:49, 9 April 2006 (UTC)[reply]

Another free program that anyone should be able to get access to is ChemSketch, but I'm not very familiar with its capabilities or suitablity for creating optimal images for Wikipedia. --Ed (Edgar181) 15:42, 9 April 2006 (UTC)[reply]

I'm quite familiar with Chemsketch. I personally find it easiest to use. Check out some of my contributions. Might not be the best, but it is presentable for many purposes. Some drawbacks (or maybe I haven't figured out how to fix it yet)

• how to distribute objects evenly (vertically or horizontally)
• how to make special reaction arrows
• how to use non-standard valence or oxidation state
• how to stop displaying the terminal -CH3 --Rifleman 82 07:38, 1 May 2006 (UTC)[reply]

To distribute objects evenly in ChemSketch, simply switch to Draw mode. Select the objects that you want to have aligned by lassoing them or by clicking objects while holding SHIFT key. Then, click appropriate vertical or horizontal buttons to align as desired or use the Object>Align Horizontally or Vertically menues. Once this is done, you can group corresponding pairs elements at both ends toghether and then use the centering tool.

Selecting special arrows in ChemSketch can be done either in Draw mode or in Structure mode. In structure mode, you can select from 12 different reaction arrow styles by hitting the Reaction Arrow button list (see small triangle in lower right of the button). Alternatively, you can also use the context sensitive menu on any drawing to change its style manually.

Setting non-standard valence in ChemSketch is done through the Structure Properties dialog box. Click the Atom Tab, and it will allow you to set not only the valence and oxidation state (i.e., charge), index position, Isotope, and numbering.

You can hide or show termial carbons using the Structure properties dialog box. Under the Common tab, clear the show terminal carbon checkbox.

It is worth mentioning that Journal styles, including those supported by ACS, can be applied to a molecule by selecting the appropriate journal from the drop down list in the Structure Properties dialog box. You can also save your own customized custom style for later reuse.

Michel Hachey 21:57, 11 October 2007 (UTC)[reply]

According to the Wikipedia, Inkscape is an open-source vector graphics editor, that has incorporated the bitmap-tracing functionality of Potrace. I will try to use this software to make chemical SVGs in conjuction with ChemSketch, ISIS, and ChemDraw, but will not have time to do it until Sunday at the earliest, and if anyone beats me to it, they are welcome to do so. Shimmin 14:37, 13 April 2006 (UTC)[reply]

I got around to it earlier than I thought I would. It works, but the results aren't as pretty as they might be. The potrace algorithm apparently doesn't "do" corners very well, and so bond tips are noticeable rounded. The extra lines of multiple bonds are lozenge-shaped. The version of Inkscape I downloaded does not appear to be able to import WMFs, either as files or through Windows copy-and-paste, but I'm still very unfamiliar with the software. Shimmin 15:31, 14 April 2006 (UTC)[reply]

It appears I end up re-iterating what someone else said above in saying this, but the SVG Factory for some reason does not convert the textual elements of an WMF along with the lines, so I when using it, I end up with chemical structure diagrams without symbols. This happens when importing from both ChemDraw and ISIS Draw. Shimmin 16:00, 14 April 2006 (UTC)[reply]

I've had some unfortunate experiences with SVG in chemical drawings. SVG is not yet prime time. Here's why. Compare these two images:

SVG:

PNG:

I created the PNG image for the Blaise ketone synthesis article. User:Scott5114 used Inkscape for SVGize it. I don't like the SVG version. The letter spacing is too narrow. The font is changed. The bond lines are too narrow.

The largest problem is that the SVG version looks really bad in IE. ~K 01:06, 18 April 2006 (UTC)[reply]

PNG is a raster format. If you start with a raster format, and go to a vector format, the quality of your image will be constrained by the quality of the algorithm used to trace the raster. This is a tricky problem for highly detailed parts of images, like fonts. I am not surprised by your above result, but it is not a fair test of the SVG format, merely of Inkscape's ability to vectorize PNGs. To do this right, you need to stay with a vector format the whole way.

I have been told that Open Office can convert WMFs (windows metafile, the format used within windows for vector drawings on the clipboard) to SVGs. I do not have Open Office installed. If anyone does, I'd be interested to see your results. Shimmin 11:39, 18 April 2006 (UTC)[reply]

The largest problem with SVGs on Wikipedia are the fonts. The tool used to convert SVGs to PNG for display purposes seems to lack many fonts or at least renders them horribly in some cases. For text that needs to stay at a fixed position, needs a specific size and font, I'd suggest to convert the text to outlines. That way the textual information is lost but display fidelity is preserved which seems to be paramount for WP purposes for me. Images aren't searched for content anyway (at least not yet). --Johannes Rössel (talk) 09:09, 18 June 2008 (UTC)[reply]

After my persistent harrassing of the poor folks who so generously give their time answering dumb questions from technically challenged people like me, I got a clear answer:

"The size (in bytes) of a thumbnail image does not depend on the size (in bytes, or in pixels) of the original full-size image. Period. The only exceptions are if

1. the thumbnail would be bigger (in pixels) or the same size as the original, or
2. the original image is so huge (something like over 10000 pixels across) that the MediaWiki image resizing code can't handle it."

So we can make and upload images pretty much as big as we want. But how big do we want them to be? Please add to the discussion above. Walkerma 05:07, 14 April 2006 (UTC)[reply]

I will take silence to mean that consensus has been reached on using ACS settings for structure drawing. I (or one of you) will update the style guide accordingly soon (I have a lab to go to now!). If you have any problems with using ACS settings, please post something on this page so we can fix it.

What should we do for the next phase? I wonder if we should focus our attentions on laying down standards for 3D drawings. For this we need to bring in the Jmol group. Our focus will be on settings (colours, lengths etc.) rather than implementation. For myself, I'd be happy to see the Jmol folks give us their suggestions, and let us discuss how this fits with other 3D formats. But I know very little about 3D images, I rarely use them. Thoughts, anyone? Walkerma 16:47, 27 April 2006 (UTC)[reply]

I am not an expert on Jmol although I have put Jmol images into several web pages. I think we just have to see how it goes. Some things such as colours are fixed in the Jmol applet. If we do not like them, we discuss it on the Jmol e-mail list and get changes to the applet. That would flow through to what we do here once we install the new version of Jmol on the WP server. We need to also detemine the data file types allowed. On the Jmol wiki, only *.cml, *.pdb and *.xyz are allowed. --Bduke 21:47, 27 April 2006 (UTC)[reply]

Colours are not really fixed in the Jmol applet: all colours can be changed through script commands. A default script can be defined when installing the Jmol extension: it could be used to apply coherent settings (if more features like this are needed, tell me and I will add them). If you want to upload data files with the "upload file" feature of MediaWiki, you will need to decide which extensions are allowed. If you want to put data in wiki pages, deciding which extensions are allowed is not required but should still be done to ensure coherence. NicoV 16:29, 28 April 2006 (UTC)[reply]

OK, it looks like we have a quorum already! I forgot to post on WP:Chemistry last night, that's now done, I hope a few others may join us from there. I think first we need to establish what software we are using, and then start discussing simple 2D structure drawing settings. If people want to discuss other issues, though, please feel free.

Hello everyone, just dropping in as an outsider here. On Phenol red you'll see a nice chemsketch-produced diagram showing how the structure changes with increasing pH. I think the minus at the sulfate's oxygen is too small and could easily be mistaken for a dot, indicating a lone electron. (For a chemist such an unpaired electron would obviously not make any sense there, but lots of our readers are half-educated like me.) In the original-sized picture the problem is not as severe as in the scaled down version actually shown on the page.

So I'd like to respectfully submit that the powers-that-be consider increasing the fonts of the drawing software when it comes to minus-subscripts.

Best regards, AxelBoldt 17:17, 2 October 2006 (UTC)[reply]

Good point. I usually change them to Symbol font, that does the same thing. Of course if the software producers had any sense it would be automatic...! Walkerma 19:33, 2 October 2006 (UTC)[reply]

It is possible to set charge font style to Symbol, and this should provide an easy automatic solution. It is generally not advisable for software producers to break font standards by creating exceptions. An exception which is good for one person may easily be considered bad for others. In the interest of promoting consistency, consider choosing an appropriate font type that can be consistently applied without requiring exception rules. Michel Hachey 21:31, 11 October 2007 (UTC)[reply]

I have successfully created SVG structures buy creating a structure in ChemDraw, saving as .wmf and then converting to .svg using SVG factory, and then making some small change to the .svg file, which I describe bellow.

• 
  [7]circulene
• 
  [6]helicene
• 
  β-farnesene
• 
  α-farnesene

The svg files that SVG Factory does not have the svg name space in the file. This need to be manually inserted into your file. the top of your file should look something like the following (inserted text in bold)

<?xml version="1.0" encoding="iso-8859-1"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 20000303 Stylable//EN"
"http://www.w3.org/TR/2000/03/WD-SVG-20000303/DTD/svg-20000303-stylable.dtd">

<svg xmlns="http://www.w3.org/2000/svg"  width="3.0000in" height="2.9221in" viewBox="0 0 231 225">

I will investigate why the text problem is happening tomorrow.

Gary van der Merwe ^(Talk) 21:32, 9 November 2006 (UTC)[reply]

I'm having a problem with an SVG and this looks like the best place to go. I uploaded this image to the commons (an SVG made with ChemDraw and Inkscape), and I can't get the picture to show up. It opens just fine in Inkscape, so is it just my computer/configurations, or did I make a mistake in creating the image? Any help would be much appreciated. —ShadowHalo 09:13, 4 January 2007 (UTC)[reply]

I can't see it either. Why don't you reupload them? --Rifleman 82 09:32, 4 January 2007 (UTC)[reply]

I think the problem is that when you paste from ChemDraw into Inkscape, it copies the images as a PNG. Then when you upload the .svg file to the Commons, all you're uploading is a blank square and the code in the files is a reference to that PNG. As the PNG is still on your computer, and not on the Commons, all you see is a blank!

Silly ChemDraw, basically!

I could, of course, be wrong, but this is what happens in my experience.

Ben 10:21, 4 January 2007 (UTC)[reply]

Oh, okay. That makes sense. So how does one go from ChemDraw to Inkscape without creating a reference to the ChemDraw image? —ShadowHalo 03:38, 6 January 2007 (UTC)[reply]

I don't know! I wish it would work. Does anyone else know?

Ben 14:45, 6 January 2007 (UTC)[reply]

Since you're talking about images, perhaps Ben or someone can tell me why the image on the drugbox of ursodiol doesn't display properly...? --Rifleman 82 16:49, 4 January 2007 (UTC)[reply]

Not sure, sorry Rifleman. I had a look and tried it at different sizes. It displayed properly at 800+ px but not at 700px. Weird. I've never experienced that before.

Ben 17:55, 4 January 2007 (UTC)[reply]

Hi,

many of you, who are using incscape to create SVGs of chemical formulas will have noticed that subscript and other baseline elements are not correctly rendered by MediaWiki software. There exists an appropriate bug report on Mediazilla, but there are still not enough votes to get any attention for this issue. As user on German Wikipedia I just wanted to ask you to vote for this bug, so that we don't need categories like this any more. Thanks. --Taxman_de^{Talk (de)} 12:28, 26 February 2007 (UTC) [reply]

I have updated Wikipedia:WikiProject Chemistry/Structure drawing, please have a look at it. Cacycle 02:55, 21 March 2007 (UTC)[reply]

Looks good! I just need to write up the IsisDraw stuff now.... Thanks, Walkerma 04:23, 21 March 2007 (UTC)[reply]

I've notice that the numbering of R-groups is often given in subscript by editors e.g. .

I think it is more clear to give this numbering in superscript. e.g.

Subscript is usually used to indicate the number of substituents (e.g. -NR₂, might be misinterpreted to mean a Nitrogen attached to two R groups). By using superscript, any possible confusion is avoided (i.e. -NR² means a -R² group attached to Nitrogen. -- Quantockgoblin 00:46, 21 March 2007 (UTC)[reply]

Yes, superscripts get my vote every time on this. It annoys the heck out of me when I see subscripts used in JOC! I know that in some molecule editors it tends to spit out subscripts as defaults, but is there any reason we can't state that we want superscripts for cases like this? Walkerma 03:09, 24 March 2007 (UTC)[reply]

We should outlaw subscripts :-) I'll add it to Wikipedia:WikiProject Chemistry/Structure drawing. Cacycle 03:14, 24 March 2007 (UTC)[reply]

It is not advisable to make a universal rule for R-group. In this case, it is wrong to dictate a preference for either superscript or subscript because there is no consensus on this, and a rule without consensus will be ignored or broken time and time again. Consider the fact taht an atom can have a valence, a charge, index number, isotope number, and numbering. The positioning of all these values is not totally arbitrary, and yet you will find that conventions are only loosely followed in many cases. Drawing a 2D structure often forces one to position numbers differently than we would want for the sake of clarity.

In the case shown, it is a relatively simple molecule, but as soon a you get into complex cases, you will find that people will have bigger worries than the choice of superscript or subscript for the R-group.

Michel Hachey 21:52, 11 October 2007 (UTC)[reply]

Question, what do people think of chemical structure drawn in a simple linear format like this:

(for the record the above may be corrected so I'll state that the images are not drawn with sp3/sp2 hybridised atoms)

For my 2p worth, I don't like them as they are not strictly factually correct. Maybe I'm just being a stick in the mud? -- Quantockgoblin 02:06, 24 March 2007 (UTC)[reply]

Granted, these don't really indicate the bond angles, but I don't think they're actually wrong are they? If I set up my computer monitor vertically, and draw propane with a "kink" in it like so:

then I move my eye through 90 ° so it's now looking down on the structure from above, then what I would see would have all three atoms in a line, corresponding to the "ethyl group" picture shown earlier, with R = methyl. Given that this style has been widely used, and indeed is still seen in organic papers today, I think we should accept it. This is particularly true for short things like the ethyl example. I think I might usually prefer to see a bend, but to me it's not a big thing, and in some instances it can look neater to draw things in a linear fashion. What do others think? Walkerma 02:55, 24 March 2007 (UTC)[reply]

I think bond angles should be represented in pictures as close to nature as possible, that's what structure images are for. Therefore I am all for chicken wire notation. The upper image is misleading as it mixes semi-correct bond angles with completely wrong ones - it should be replaces asap. For something like R-CH₂−CH₂ we do not need images, however, this is an adequate in-text style. Cacycle 03:11, 24 March 2007 (UTC)[reply]

Walkerma, true it possible to draw molecules at very unclear angles whilst being factually correct i.e. n-hexane as a single long line, but that is not really the point. Our job is to be clear and accurate. It is true to say molecules drawn at 90 degree angles are easy to understand (and have the correct connectivity), but I'm not sure they are strictly correct. However, molecules drawn with 120 degree are clear and still easy to understand, and at least when drawn with 120 degree bond angles, there are normally a few angles (i.e. looking down the axis of a bond) where this representation is "true" i.e. the bonds appear to be at 120 degrees. I think the only excuse for the 90 degree approach is when drawing all the hydrogen atoms on a molecule, where otherwise it all becomes too congested to see what is going on. e.g.

or where in some sugar diagrams e.g.

File:D-glucose Fischer.png

where these sugars are drawn 'flat' - but that is done to convey the complex arrangement of the chiral centres (e.g. left = R and right = S etc ... probably guessed the wrong way round!)

In the end it might just be a question of style - but which ever style we pick e.g. chicken wire notation or the 90 degree approach, I think we should try and be consistent and try to encourage everyone to use the same style where they can. -- Quantockgoblin 23:01, 24 March 2007 (UTC)[reply]