Winter Project

If it helps you any our computers at work are dial up. It took me approx 45 seconds to fully load the lady bird page at 2:30 am.

On dsl at home it took approx 7 at 3pm.

I'm so glad you brought that up. I downloaded that page to my PC and found that the thumbnails currently add up to around 800Kb, which makes browsing over dial-up a real pain. Luckily there's a relatively easy fix. (hopefully John is reading this)

I noticed that over half the thumbnails still contain the EXIF data (and possibly IPTC data, embedded thumbnails, etc.) from the original image files, so I ran 'convert -strip *.jpg' on the directory and trimmed them down to around 180Kb!

For a rather extreme example, this thumbnail is 48,571 bytes. After stripping the non-image data from the file it's only 3,922 bytes.

To answer your question, I think 40-60 thumbnails isn't too bothersome on dial-up as long as they are only 2-5k each, which will hopefully be the case in the near future. I suggested this same fix to the editor at naturephotographers.net a couple years ago when I was on dial-up and it made a VERY noticeable improvement in the critique forums, which display 40 thumbnails per page.

OK, exif info has been stripped from the thumbnails. Or, I should say, is being stripped since BugGuide dynamically generates thumbnails to a cache as they are requested. So BugGuide will be a bit slower than normal for the next day or so while it regenerates all 58,000+ thumbnails, but the decrease in bandwidth should be sizable and especially felt by those on dialup.

Looks like this isn't happening anymore. I just noticed a few thumbnails as large as 60k, which is just a little smaller than most of my uploads!

Thumbnails in ID Request, the Lady Bird family page, and everywhere else for that matter, are loading so quickly now! Anytime the data to be transferred for a connection is smaller than the TCP window size, everything just snaps onto the screen.

Thanks John!

I second that motion, I'm at work and it feels like i'm surfing on dsl!.. Great suggestion Jay, and thanks to John for implementing this function!!

Will this function also integrate to any newly uploaded images? In other words, will all exif automaticlly be stripped from any all images, or is this a function that will be ran once a month or so to process any new images since the last "stripping"?

BugGuide has dynamic caching, which means if there is no appropriately sized image available (e.g., you just uploaded an original image) one will be generated and stored.

exif is retained on regular images and is only stripped on thumbnails.

Thanks, Jay. That's very helpful. I'll look into it.

I had no idea such things were possible. 45 seconds is a long time to wait for a page to load. I'll email John and see if he can do some "stripping" on the pages before I do any more. Thanks for the help.

of putting a thumb of different species under identification. pretty good idea

That looks fantastic Lynette.

Looks good Lynette, nice work.

I'll add that to my list. It would definitely help me out. I could learn them doing it and then have them for reference later on. :)

Actually I think you should tag all 25 as rep. as suggested. The reason I would prefer you don't select only 12 is becuase when I browse it automaticlly shows 12 random representitive images per page. When you only have 12 selected it's the same 12 each time.

I think alot of people besides myself look for insects by selecting a close genus and repetedly click browse to load new random images until you get one that matches your insect closer and closer until you find what your searching for.

Therefore to make "browsing" for an insect better we should have at least one from each species marked as a rep image. That away there is plenty of random images for the page to choose from. Of course upkeep will be a hassle, maybe the representitive tag should be OFF by default an leave it up to editors to change the tag when a better representation is found?

I think I will leave that field alone and show the other images in the body of the Info page as Phillip is doing with the Coleoptera page.

Choosing representative images for a guide page only affects that guide page. If you were to select 12 images that truly represent the lady beetles, when someone Browsed on beetles they would eventually browse their way to that page, recognize their animal as a lady beetle from the twelve wonderful images you have selected, click once again on browse, and have all the lady beetle genera available for view. We see no harm in deciding that good images of a variety of lady beetles show up on that browse page. It is much better than having random small inages or body parts or discovery locations show up. Or, and this is especially true for lady beetles because probably 80% of the images in the guide are Asian Multicolored, having a browse page show up with most of the images belonging to a single species. And the only other alternative is to go through every image, edit them, and turn off representative. Try browsing spiders on a search for Argiope aurantia. You will not know we selected the twelve images for aurantia till you get to species level. Until then any image is just as likely to appear. But when you get to aurantia, you will know that here is your animal because, we think, we have selected a range of images that well represent the species.

with what you are saying, but I know quite a few people use that refresh button just to see what will come up next. I guess they are using it as a mini search engine, (sometimes just looking to see how many images are their own). While this isn't a very effective way to search, I'm not sure I want to take away their ability to continue to use it that way. Perhaps the better way to get rid of the poor images will be to go through the images and uncheck the representative box or frass them?

you have decided the other process is better, but on the chance that we haven't really been clear enough, we'll try again. The only page in Browse mode that would be affected by your selecting important, high quality images as represenatative of the Coccinellidae would be page four of the "Superfamily Cucujoidea - Flower, Flat Bark and Ladybird Beetles". That is the only Browse page where the images of lady beetles would not change when Refresh was used. On every other page in Browse mode lady beetle images would change when refreshed. It is only one step immediately higher in the taxonomy that is affected by selected images on a guide page. We think this is a valuable tool to use to really allow people an oportunity to see the makeup, the breadth, of a species, a family, an order even.
Wouldn't it be great, for example, if the Neuroptera Browse when viewed from Pterygota showed one excellent image each of alderfly, dobsonfly, fishfly, dark fishfly, snakefly, green and brown lacewing, owlfly, mantidfly, antlion etc. The way it is now you could get three fishflies and a couple antlions and never have a clue what owlflies look like.
Anyway, we certainly trust your judgment, but just wanted to say that we think selection of representative images is a powerful and underused opportunity on BugGuide. Can't wait to see what you end up doing this winter. Whatever it is, we already know it will be a great addition to BugGuide!

John and Jane,

I must appologize for the confusion. I believe this miscommunication was really my fault. I did not poperly express myself and using, "browsing" was a bad term for what I ment. Sorry if this comes out long but I should explain.. The thing which caught me was in the origional post it was stated:
>I would like to see one representative image from each genus listed.

I thought that should be edited to say one image from each species in the guide. I don't think that would effect search time, but I do see how it could effect the browse button. For instance, When I find a new moth I usually don't use the browse button to drill down closer and closer to species. For me there are too many similar families and too many paths to get lost down. I ususally find my image by going to the moths graphic which automaticlly selects tge guide pages info tab, I look though images there and if I don't see anything close to mine then I refresh the page to load a new set of random images. Once I see one that's closer I will go to that family, then genus and so on working my way down, always clicking refresh until I find what I'm looking for.

Now to use your example, If I go to the Neuroptera info page, I get a bunch of random images of anything tagged Neuroptera, and everything else below it all the way down to the species. I could concevably have an unidentified Neuroptera right next to a Summer Fishfly.

If I go to the browse tab it uses the next levels of Neuroptera. In this case Suborders: Megaloptera, Raphidiodea, and Planipennia. Browse tab shows me 10 or so random images from each of those three groups or anything under it in the tree. In the Megaloptera section I could have an unidentified Megaloptera next to a fishfly, but not a snakefly, becuse that would be in the Raphidiodea section.

Now I asked myself how would changing the number of representitives per speices genus or family, effect the info or browsing?

Well if there wasn't at least one representitive per species my chances of finding my insect are zero the way I currentlly use bugguide, since a non rep images would never show up in browse or info. Once I was close to species level I would eventually have to go to the Images tab to find my insect. Personaly I wouldn't like that.

If there was many images marked rep per species my method of browsing could contain too many shots of one insect even at family level. Thus making it unlikely I would see my insect with out having to refresh many times. Even more so if we only had one other rep image in the guide of what I was looking for. It would be statisticlly against me. Frankly that is kind of how things are becoming now. Personally I find this a little frustrating also.

The key is finding the happy medium, How many is too many representitives per species, and is it concevably valuable to have at least one per?

With the large numbers of images submitted, and the many variations of species, I don't see the harm in having the rep tag based on species variation, for instance on species which male and female are closely matched have one for that species, where there is a noticable difference have two, one male and one female, where there is a sigificant pattern differences by sex and region, maybe even have four(two of each sex). I think in species with a very high rate of difference (asian ladybeetles, P. audax....) we should limit the those types to under 10 representitives per species.

I don't think 10 for a few select species is too much?

Obviouslly any representitive image would ideally be an image that is well composed, lighted, focused, and show the insect in a true likeness. Unless it's the only image of that species, then a less then favoriable image should be acceptable until a new better image is posted. Then then whomever moves the new image over to the guide page should be responsible for unchecking the rep box on the older image of less quality.

I also agree that any editor when browsing should be on the lookout for images which shouldn't be representitives and unchecking the box when we see bad images of common species while browsing. That would do a lot to help this problem.

I do think each guide pages info tab could be edited to show some of the representitives below it, in addition to the random images automaticlly placed at the top of each info page. There is nothing wrong with including images in the description for that guide page, I don't see any reason why that description shouldn't include some choice images from selected species further down the tree. Where to cap that number i'm not sure, maybe 12? Any more and you'd have too many photos to load per guide page and that could really slow up browsing if each guide page contained a bunch of extra images in the description section...

Really sorry about the length...
Gehan G.

I probably do not fully understand what I am going to do, and it may change once I get going. I hope my indecision at this point isn't frustrating you. :) I DO totally agree with your thoughts on having good and diverse representative images. I know that search time would be drastically reduced. Thanks for the help!