oh, that's a very good point. thanks. first thought is that relying less on variations in hue and using more variations in intensity would help. hmmm. going to experiment. thanks again.
You cannot solve colorblind problems with these changes alone. The best way, is to change shape.
There are different types of colorblindness. The worst being the one where you are totally unable to discern colors. Hue and Value alone won't solve the problem.
For example, at first I didn't see the difference between "Alice" and "Bob" on your about page. I had to look closely to see it. And I'm only mildly color-blind.
I suspect that by using the word "intensity" to describe what he's going to change, andrewcooke is well aware of this already :)
For everyone else: there's a color model that is occasionally used in computer vision called Hue Saturation Intensity, which pretty much describes the theory of color that andrewcooke and bornon5 are talking about.
Yes... I mixed up the terms. Thanks for spotting that.
While I have never fooled around with it, I think using the LAB colorspace could be interesting as well. As I remember from my psy course, the two most common forms of colorblindness are red/green and yellow/blue.
Given that the LAB colorspace models these exact axes, it could be interesting for something like this.
But as I said, I have not yet investigated a lot in this direction.
Well, I don't see how this can be useful. Human brains are terrible in remembering faces, but not in remembering colors. Actually, comparing the icons (to find out if they are the same or not) is quite hard. Have you tried it?
it's not intended to replace photographs. in fact, one thing i would hope to do in the future is allow people to upload photographs. instead, it's meant to replace the auto-generated images that already exist that use IP addresses (like on stack-overflow, for example). those are really ugly; mine are much nicer (and more flexible).
while i was developing this approach i tried lots of ideas. it's actually quite hard to get images with (1) a wide range, to avoid "collisions", (2) attractive designs, and (3) coherent forms (that don't look like a random mess, making them easier to remember). one of the things that helps a lot in making these more distinctive/individual/memorable is the diagonal symmetry. another is the generally consistent hues within any one icon.
I did something similar for my users' default avatars, randomly picking an image from a directory of a few thousand images I gathered and renamed sequentially. I just scoured deviantart.com for texture icon packs (usually 100x100px) that people seem to make for some reason, and threw them all in there. I spent a night quickly flicking through them and deleting anything exceedingly boring, too effeminate (majority pink/purple), or just lame. It hasn't gone public though, so I haven't experienced reactions yet.
> [...] those are really ugly; mine are much nicer (and more flexible).
This is a matter of personal preference. The more people you ask, the more opinions you will get. I could chip in mine, but that would serve no purpose.
You could see statements like these as a form of libel/defamation. While the statements are not necessarily false, they may cast a negative light on whomever is generating these images (in this case, gravatar.com).
I would try to avoid these statements, and simply go ahead with your project. And let the people decide for themselves which one they prefer.
He's being honest. The existing solutions aren't any good, so he made one that is better.
There are thousands or millions of FOSS projects that were started because the existing ones are bad, and they've all said that they were started because the existing ones are bad. It's just the way it works.
Now, if both he and his competitors were serious companies with a handful of employees that were charging money, then it would be less appropriate. But they aren't. His project was created because the competition's projects are ugly, and he should be able to say that.
> it's meant to replace the auto-generated images that already exist that use IP addresses
I think you need to emphasise this point more in your 'about' page. I skimmed through your about page and came away with the thought 'I don't think I'd want to replace my personalised avatar with something like this - it's much harder to identify me through this than a photograph or illustration'. But as a replacement for a generic avatar - that makes sense!
apparently so. thanks for that. the images are so small that they are also cached in the datastore (they are cached at every level - in html, in memcache, and in the datastore), so this is not a big deal at the moment (with just a hundred or so images), but i need to disable this in the future.
I've been working on this on weekends for a while now; it's now stable enough for some early adopting... Please report any bugs and feel free to ask questions. You can also contact me at andrew@parti.cl. Thanks.
When it was first announced it seemed terrible - the estimate for even the unused development site was significant. But then I tracked that down to some dumb coding of my own (a much-too-frequent cron-job). With that fixed, the site is pretty efficient (which is by design - it's supposed to scale well, after all). So I'm hoping it will work out OK. This is very much (I think) the kind of site that should work well on GAE - there's a lot of scope for caching.
And this is one motivation for posting here - I want to see how well it handles higher traffic. Hopefully from that I can work out what a decent amount to charge would be (originally I was hoping for as low as a dollar a month a feed, but the changes may push that higher - perhaps significantly so).
yes. but i'm only planning to charge what is needed to cover the costs. if those are reasonable then i think some people would be willing to pay for something a bit more attractive. if not, well, at least i tried.
(i'm not adverse to making money from this, if it becomes wildly popular, but it's not my main motivation. i did the ground work years ago and then posted a link to my notes in the thread on the robots approach here. someone contacted me about that, asking if i they could use it. i suggested a service, since i wanted to learn gae, and they thought that was a good idea...)
The rub is that you already have Gravatar, which is 100% free, offers several options, and is easy as cake to implement... What compelling reason would someone have to use this service (and pay for it) that they cannot already get at Gravatar? i.e. what problem are you solving, or what underserved niche of Gravatar's customer base are you targeting, and how are they better served?
I'm not trying to be rude, so don't take this the wrong way, but what is the point of this? What problem/desire does it address? I'm just not understanding.
That's a very attractive site, and I really find the icons attractive, too (though I imagine they'll be unhelpful for people with difficulty discerning colors).
I thought I'd let you know that you have "DEBUG = True in your Django settings file"[1], which reveals your various routes. This might be unintentional.
This sort of mosaic-y, rigidly geometric style would only fit in with certain website designs, and the colors in the examples are all extremely saturated. I think having to sort through these would be really exhausting on the eyes.
I hate to say it, but it's a common rule of thumb in graphic design that people like to see other people. Faces. That's what user icons tend to be, because it's what works.
see my answer elsewhere (and i'll update the site tomorrow) - this is not to replace photos, but an alternative to other auto-generated icons.
also, if anyone wants less saturated images, or any other variation, please contact me and i'll see what i can do. i'm happy to iterate this (for example, things will likely be changing to address the colourblind issue raised elsewhere).
I prefer monsters or robots. Those abstract avatars tend to exert strain on my brain. But I think we need more options for this kind of service, so kudos for making it.
If you make such a service, I think you should put some research into what kind of thing the brain is good at recognizing. For example I think brains are almost hardwired for facial recognition (it is sort of a miracle that we can tell people apart so easily - try it with people of other ethnicity to get a feel for the problem). Not so much for abstract art I suspect.
I like the minimalist design of the site, and I think your idea is the most visually appealing of all the default avatar options out there.
Some ideas:
Landing page:
- Add at least a quick blurb / product explanation to give first-time-visitors a quick, visual idea of what the product is (maybe even an alice/bob style demo conversation).
- Have more descriptive action choices "ready? start now for free" and "Learn more" instead of just "about/configure". About pages are somewhat ambiguous and can sometimes just be about the company / the people behind it instead of the product explanation, you're basically giving first time users nothing and hoping curiosity prevents them from bouncing. I'd say give them at least a bit of a hook to get them to venture further.
About page:
- Enlarge and really highlight the alice/bob chat box -- this is the main function and selling point of your app and is the first/only time it's visible, and it's kind of tiny.
- Cleanup the available styles example a bit -- it seems like you're showing every possible style available which when placed all together next to each other looks downright busy. I would pick maybe 3 or 4 icons that showcase the important variations available (size, cell-margin, color variation, etc.) and showcase only those, maybe with alice/bob style examples to show what they would look like in the flow of a conversation versus directly next to each other.
Configure page:
- Change the phrase "New users receive a free credit to evaluate the site" to something like "Try the service out for free" or "Evaluate the service for free" Explaining the whole credit thing on the sign-up page is way too much for someone just looking to test the waters in my opinion.
- Simplify or better organize the small print section, maybe separate out into sub-sections: explanation of how billing works, what you can expect from particl as a service, etc. At this stage in the signup process, you haven't necessarily sold the customer yet, don't lose them with scary large blocks of confusing text, especially if it's unnecessary. I'd either make it digestible or move it to a more appropriate section of the site and link to it.
General ideas for the business model -- I frankly don't see people footing a monthly bill for -just- visually attractive default user icons. I think for this to work, you have to add something else into the mix, here's what I've thought of:
- Solve the color-blind / accessibility issue and highlight that feature
- Offer a turnkey service for site owners to drop in to their own server, maybe even a complete commenting solution if possible.
- Offer a hosted complete commenting solution similar to disqus but with a premium and minimalist design.
- Offer tiers of service based on number of registered users, so sites start out with your service because it's free, and stick with it because it works, it's already in place, has good design, and is affordable.
- This is a bit hokie, but maybe offer a way to hash something more meaningful than a username, like say have a "personality quiz" that generates a badge for you based on your answers to questions.
- Have the user icons form a photo mosaic of something - maybe the site's logo?. I realize this would be difficult due to the simplistic color palette most logos have, but you might be able to come up with something
for this.
- Sell directly to the avatar-holder. I think there is a bit of cognitive dissonance between people who won't upload a photo to their account where they feel obligated to the community to put something there and want to customize their online persona, but don't want to reveal their identity to the internet. I think you could capitalize on this by offering a way for people to assuage their "guilt" over not contributing by maybe offering a service where you could donate to a cause, and in exchange get an exclusive avatar that has to do with the cause they've contributed. This would let people 1) feel less guilty about not "joining" in on the community 2) make their online interactions more personalized safely 3) spread awareness for a cause they care about 4) give those who care about it a little badge of honor. Maybe combine this idea with the mosaic idea and let people have a "piece" of whatever image represents whatever charity.
thanks for all that. will think them all over. have been very surprised by the number of responses here (just woke up).
the single level of service came from before the GAE pricing update when i was aiming at a dollar a month. do you think even that is too high for people to consider?
No problem - just what came to mind when looking at it. For any things you do consider changing, be sure you are keeping good track of the relevant statistics to help you know what is working and what isn't.
I don't think $1 a month is too high; I think the main difficulty will be just getting the consumer to overcome the mental friction inherit in any online purchase, especially subscription purchases. You're going to have to convince people it's worth getting their wallet out for in the first place, and after that I would guess you could price it anywhere between $1 and $10 and it wouldn't drastically change how many people convert. That's at least given the current state of the product - if you can demonstrate more value, then you may have an easier time convincing people to spend their money on it.
I still think your best bet is to offer a free plan to small userbases, and then charge them once they grow beyond a given size. You could reassure them that they won't lose their images or have their account held hostage if they don't upgrade, it will just be frozen at the max limit of user icons. I would also offer 2 upgrade plans: a basic plan which includes 1 or 2 styles/sizes of icon, and a premium plan with all the customization options. The goal of this would be to make the basic plan more attractive while allowing those who need to customize to do so. Another idea is to pin the free userbase option as a "startup" option, and sell it as a gesture of goodwill towards entrepreneurs vs. a free/paid plan.
I really don't know much about GAE, but do they have some way to sell this as a GAE-compatible plug-in to other App Engine users? That might be a good route to take if they do.
