(This is the third of my series on indie covers. For the previous two, look at last week and the week before, on Wednesday.)
If I had a dime for every time a newby writer sends me a cover with either a picture of some trees, or a bridge or something or the generic Amazon cover with some words, while the novel is “science fiction” I’d have… an awful lot of dimes. Possibly enough to buy out Elon Musk, that’s how many dimes I’d have.
When I ping these misguided children (some of whom are older than I, I get told that “But that’s the art I could get by taking a picture. And there are some trees in my novel!)
Alternately I get told “I don’t want to be sued.”
Oh, mind you, I don’t particularly want to be sued, either. But there is art you can get rights to, for not very much. (I haven’t done it in a while, since I’ve gone off the reservation with either free or my own renders then altered (Or a combination. As a note, if you render, DO try to get the updated models, not vicky 4, not unless you’re doing some amazing after-processing. No, seriously. There’s a difference. And it always needs post processing, otherwise you get … plastic dolls, by the positioning, even when the face avoids the uncanny valley. I’m getting better. Not perfect, but better.) Back when I bought ready-made art it went for about $10 a picture for my purposes. I don’t know now. I know you can buy bundles, and I do when I stumble on them (mostly for effects and brushes, TBH, but I don’t have the names handy. I’ll try to have them next time.
When I bought art, my favorite was dreamstime. But there are about a million others, give or take. I just liked them, because they had the best search abilities, or at least the one I’d honed until I could bring up precisely what I wanted on the right search string.
Speaking of words and art, just in the last week art bots have burst into my consciousness. I’m sure they’ve existed for a long time. I remember when everyone was turning their picture into a 17th century portrait. (I would have too, but for some reason the bot decided I was a guy. Sigh.)
Midjourneybot, available only on twitter does amazing “pulpy” art. I tried it somewhat, and will buy a subscription, after the novel is done (No, it’s not done. Shush. My muse is a sadist.) The style of art is not appropriate for science fiction, and probably not for fantasy, but I think I can do two things with it: Create the sort of impressionistic not too detailed backgrounds I want for my scifi covers, AND do historic mysteries. (Figuring out how to subscribe is the problem, just now, partly because my discord is WEIRD.)

Again, my acquaintance with these bots is very minimal, because I’ve been living in a hole with a novel that’s treating me very badly.
However, I’m familiar with websites that have free photos. The photo for this post, for instance, is from Pexels.
If you click on the notice, this is the notice:

Now I don’t often use pictures from Pexels for covers, particularly of science fiction, and particularly if they have people in them. Because you can’t be sure that the photographer got a model release, and though you’re unlikely to be sued, it could happen. More importantly, the cover is too photo-realistic for a cover.
So, suppose you run it through some filters, (I own filter forge professional, which can also be installed as an add on to photoshop (which I don’t have) and you get something like this:

However, mostly, if I need a cover in a hurry, I hit pixabay (you don’t even need to credit the artist, though it hurts you nothing.)
This is the results on the arcane search “science fiction cover.”
Area most of them really good covers? No. Which is why you should do your research and look at covers in the field. For one most stock sites are iffy on the concept of science fiction versus fantasy.
BUT this one would be better as a generic cover for space opera than a picture of some trees and a park bench.
As is this, if you remove that star wars thing from the front, which ain’t hard, since the fund is black
So, the second, with automatic photo fix, for a bit of clarity and the weird ship gone looks somewhat retro, frankly, like one of the GOOD SF covers of the 60s. Great for someone who writes a bit retro-future, maybe. And it’s FREE.
But it gets better. Suppose instead of being a complete dorkmeister, you go more specific in your search, and search for science fiction woman (and select illustrations)
Time travel isn’t as productive, and I’d prefer a human figure here, but eh, it wouldn’t hurt a book, either:
This one is a better time travel cover, but it was hiding under “magic” (and needs proportions fixed. This isn’t smashwords.)


Also under magic, I would not disdain having this as a cover of a medieval fantasy with dragons. I’d probably recenter it, use the dragons for the front cover, and move the rest around to the back for paperback print. Or you know, just turn it around.
And this one would be great for a medieval fantasy. Pexels is also great for urban fantasy background:
BUT all of these are free pictures, just laying around, waiting for you to use. You don’t need to have no picture, or a road side travel-log like picture.
So, go for it, and have fun. Go poke around. For that matter I know there are other free picture sites.
One caveat: always run free pictures by tineye or another reverse image search, because they allow people to upload without check, which is why if you search for medieval woman you get a picture of Anne Boleyn from the Tudors.
If you don’t watch a ton of tv, you might not realize it, so always search.
As always, I look for you guys to add other sites and things of interest in the comments.
Next Week: The great manipulator: Tips tricks, etc, on manipulating images.
Notes on Midjourney for when the novel is done 🙂
I found subscribing to midjourney easier on desktop than mobile. They take credit cards through Stripe, and warn against some scammer impersonating them who claims to take paypal.
The best deal for people who like to tinker endlessly with it seems to be the $30 a month plan. It gives you access to the /relax command, which basically tells the /bot that this isn’t a top priority render, and basically allows you unlimited pictures in /relax mode. Can’t do maximum upscale in relax mode (ie, largest clearest renders), and have a finite amount of render minutes per month in /fast mode, where maximum upscaling is possible.
You can specify an art style, like octane render (which has a similar vibe to Lux/Iray renders from Daz Studio) or different artists. “by alma tadema” was enough verbiage to get it to try to emulate Sir Lawrence Alma-Tadema, although I did better when I was adding other Alma-Tadema compatible words like warrior or gladiator. Also recognized Remington and Mucha by their last names, and Yosh itaka Amano (the first part of the name is one word, but some comment platforms don’t like it, so I broke it into two) by full name.
Getting even slightly plausible full-length and mid-length humanoids out of it is a chore, but its spaceships and starfields struck me as more or less on par with what you find on actual covers.
For celebrity fan art and probably for historical figures we have actual images of, it does better with people who are either extremely popular and heavily photographed, or people whose faces lend themselves to caricature. I ordered the main characters from the Good, the Bad and the Ugly as pirates (results on my blog, as a Weird Wednesday post) and it took a lot more effort (and a bit of postwork) to get youngish, handsome Clint Eastwood out of Midjourney than it did for his more quirky-looking colleagues.
Forgot to add: it can do faces okayish (as in: best case results just need basic painting over or liquify for postwork) in closeups, if you run enough variations, but it really struggles with faces in mid-length and full length shots of humans. I’m going to try compositing a headshot and long view of pics I got from one prompt at some point when I have time.
I’ve gotten absurd inhuman faces and fairly realistic ones. But I lack the midjourney-fu to get anything like what I want.
There was a reference somewhere in the documentation to testing your terms in google images before using them in prompts, because that was what midjourney was going to be referencing when it builds its responses to the prompts. Once I understood that I need to approach it more or less the same way I would an image search, I feel like I got better results.
There’s some inexpensive sites, too, like selfpubbookcovers and such.
But on the AI, here’s a rabbit hole, their community show case!
https://www.midjourney.com/showcase/
What’s great about selfpubbookcovers is the way it has you do your own text layout. Very educational about what works and what doesn’t, for anyone with a basic sense of visual aesthetics.
I confess, I lean on selfpubbookcovers dot com. The prices are reasonable, their terms are still pretty good, and they have what works for the “series look” of the Familiars and Merchant series. However, you have to be careful with keyword searches, because the artists do not always use the keywords I’d use (a Snowy Owl is not a hawk and vice versa), and they have lots and lots of unfiltered photo images. If you do romance, oh boy, they are fantastic. Thrillers? Lots of options. Urban fantasy that is NOT paranormal romance? You have to get creative. They provide fonts that are licensed for commercial use (a good thing), and you have to give the artist proper attribution. They send you a link and instructions, so it’s not that hard.
and then Sarah can run a guest post about creating drop shadows…
… I’ve lost it, I think? If I got it, which I dont’ remember. You have to understand my mind is made of cheese and has been for the lat 2 years.
lost it again?
Drop shadows for what? For text, it’s very easy in Photoshop (I haven’t worked with GIMP enough to give advice on that yet.) You look in the Layers menu, select the layer you want to fuss with, then click the little “fx” at the bottom. This will give you a whole lot of layer effects to play with, including drop shadow.
I always recommend using “stroke” on cover text. A simple black or white stroke will help your text distinguish itself from the background.
For figures.
Sometimes stroke helps, sometimes it doesn’t.
I particularly note that your cover ain’t done until you have put the title on it. Especially when you could, on one hand, have “The Maze, the Manor, and the Unicorn” and on the other, “Lifestone.”
BTW, short titles are easier to work with on covers. Among many other considerations, in particular that they can signal genre quite hard. They will not overcome contradictory covers, perhaps, but the more genre one is, the more generic the other can be.
Since Midjourney keeps coming up, it seems like I would need a Discord account to use it. However, Cedar’s post was the first I’d ever heard of Discord. Can you tell me more? Am I the only person who doesn’t deal with this thing? Should I?
https://discord.com/
is a gaming chat program, mostly typing, some rather decent low-demand audio, and last time I tried it not great video chat.
It makes it easy to share pictures, it’s free (with an option to subscribe to send larger attachments, I don’t because they’re a little too dear), and if you use a burner email it’s fairly anonymous.
There’s been a few possible jabs towards trying to audit content, but they’ve been in situations where there may indeed have been illegal activity going on, or they may have just Not Wanted To Deal With It.
(Not like someone is going to say “yes I was organizing straw purchases, muwahhahah!” or similar.)
It’s commonly used for both video gamer groups and tabletop gaming that’s being done remotely, though it’s expanded to schools and other types of groups since kung flu.
It is entirely possible to make an account and use it for *nothing but* a subscription to the picture program, although I would suggest you get involved in some of the other fannish servers if for no other reason than the learning curve. IF you are on MeWe, I believe Cedar can get you an invite to a server where several folks are working on this stuff.
Book Club With Spikes invite, for you and any MGC commenters: https://discord.gg/U8FKrfAa
I managed to get as far as getting an account, but I can’t seem to get into the book club (accepting the invite gives me a “macOS does not recognize websites starting with discord:” error.
You are in, not sure how to help with making it so you can see it, though.
Try this?
https://discord.com/channels/@me
Interesting comment on Vickie 4. Several years ago (NLT 2009, but more likely ’07, I think), I found a Baen title on the city library’s New Books shelf with a SciFantasy-themed cover populated with several iterations of Mike3 in space armor (default shape and colors) and a couple of Mil-Dragon2s with their default textures. I don’t recall anything else about the book, becauze the cover was (from where I stood) so phoned-in that it got a hard pass, even with a mental allowance that the MS inside might be above-average. (For those wondering, the “Michael 3,” “Victoria 4,” and “Millennium Dragon 2” figures are 3D models by Daz3D, for use in their DAZ|Studio software or the competing Poser program, now published by “Creative Labs” I think.)
(If anyone else here uses Daz|Studio and wants to avoid the Uncanny Valley look of most 3D renders– especially if you’re looking for a comic or pulp-mag feel, email me at josh dot griffing on Gmail and refer to this column. I used to sell a CMYK-halftone shader called “Skunkville Noir” for DS4’s 3Delight renderer, under the handle “Eustace Scrubb,” and I can send you a copy if you need one.)
Poser was from creative labs at some point, fractal design before that, currently Bondware inc whomever that is.
Parent co. for Renderosity.com. I think they now hold the record (at 3) for most major builds of Poser under the same corporate banner. But itbeen a long time since I did much with Poser shaders, and mine are Daz|Studio-only.
Ages ago, i reviewed the Poser Pro PAck for a website