Selfies, video selfies, & avatars

The following are general sections on selfies - for model-specific help, look under Selfie Guides section of Help.

Prompting#

To create a selfie, open your Kindroid's gallery with the image button at the top of the chat, tap Generate Photo Selfie and give the AI a prompt. Prompting describes what should be included in the image, which means you should include attributes like location, scenery, mood, actions, etc. Note that the selfies engine does not know names, nor context on your Kindroid, so names will not work very well.

If you're getting started, we suggest you to try the Enhance button next to the prompt box in single & group selfies, to see how you can turn a simple prompt into a more elaborate prompt that can create more detailed images!

Prompt example:

• Walking in the park on a sunny day, smiling, with a few dogs

Aspect ratio and orientation#

You can adjust the orientation of the output image under Dimensions in photo selfies, group selfies and autoselfies (Odyssey or newer selfie models), as well as in video selfies when you upload your own first frame. This lets you play with composition, as the AI will be aware of the aspect ratio selected. For example, using portrait will allow you to get much more full body shots, while landscape might lend itself great to groups of 3 Kindroids in a setting. Square (1:1) is the default orientation, with portrait (2:3) and landscape (3:2) available to select. Edits made with Edit Selfie (inpainting) keep the original selfie's orientation.

Enhance selfie prompt#

Next to the selfie prompt boxes in single selfies, the overall prompt in group selfies and the motion prompt in video selfies, there is an Enhance button (subscriber only). You can type in what you want, and the enhance feature will make your prompt more descriptive and vivid for better interpretation by the selfies engine. Video selfies on V3 and V4 also have an Auto Enhance Prompt toggle that enhances your motion prompt automatically when you generate.

Video selfies#

You can also make video selfies by supplying a first frame image and an optional prompt. In your Kindroid's gallery, switch to the Videos tab and tap Generate Video Selfie to upload a Seed Image, or open a selfie and choose Actions > Video Selfie to use it as the first frame. For best results, the frame selected should be clear, and the motion prompt in cinematic language.

Picking a great first frame is also important - the video selfie will make best effort to keep identity/faces consistent, but the clearer the first frame, as well as the more dynamic room it has for motion, the better the result may be.

Video selfies are a subscriber feature and can be made with either standard selfie credits or paid selfie credits - using paid will cost less. On the latest V4 video model, a video of up to 8 seconds costs 5 standard or 2 paid credits, and a longer one (up to 15 seconds) costs 10 standard or 4 paid credits. Photo and video selfies share the same queue delay (see Queue delay below): a video made with standard credits adds more delay than one made with paid credits, and paid videos still wait out any delay you've already built up. See more at In-App Purchases.

The Video Selfie screen also lets you set the Video Duration (8 or 15 seconds on V4, 10 or 20 seconds on V3), add optional Audio (an MP3 or WAV that sets the video's speech, timing and length), and turn Generate Audio on or off. Under Additional controls you can change the Video model for a single video: V4 is the latest, V3 videos of 10 seconds or less can be extended by 10 seconds with Actions > Extend Video, and V2 is the legacy 5-second model.

Each Kin also has a default video model, which you can set in their settings under Models > Video Model. V4 (Latest) always uses the newest video model, or you can pick V4, V3 or V2 to stay on one version. Changing the Video model on the Video Selfie screen only affects that one video.

You can choose to use the SFW or NSFW engine of videos on web only - apps will only use the SFW engine. NSFW engine may be better for certain prompts/frames, but otherwise they will be very similar and the quality differences will be minimal.

Video selfies also support seeds (separate from the Seed Image first frame) - see Image/video seeds below for more on using them.

For best results, keep the motion prompt succinct and focused on just 1-2 key details such as the main subject's action or motion, and 1-2 key details to include in the video. Too much detail or description will confuse the engine and can result in more artifacts.

Group selfies#

You can open the menu for group selfies from a groupchat with the image button at the header (or with the group selfie button next to Generate Photo Selfie in an individual Kindroid's gallery). Group selfies are a subscriber only feature, and each group selfie once finished will appear in each participating Kindroid's gallery.

Group selfies lets you put 2 or 3 Kindroids of the same photoreal/anime style and composite them together. In each of those slots you can put in a Kindroid avatar, or your own user avatar (set on your active persona in Profile > Personas, with an Appearance description). Under Additional controls you can upload a pose reference (except on Palette) or use the default, and on Canon and Odyssey you can also add a style reference, just like single selfies. On Atelier, a group selfie costs 1.5 credits for 2 people and 2 credits for 3.

Under Include Kins, pick your 2 or 3 participants. Their left-to-right order in the carousel is how they're framed in the selfie. To change it, use the arrows next to each avatar to move them left or right.

Group prompts are made of an Overall Prompt and optional prompts for each participant. Tap Individual prompts to show a prompt box for each one, and Hide prompts to close them. Individual prompts are best kept concise and are useful for the clothes, mood and other details of one Kindroid. The overall prompt sets the style, setting and backdrop, and can be more elaborate.

Additional controls also has Selfie version, which is picked automatically based on your participants. You can change it for just this selfie without changing each Kindroid's default. If you upload a pose reference, it should match the number of participants and their order. When you add a style reference, Style weight sets how strongly it shapes the look. On Canon with 2 participants and the default pose, Pose strictness sets how closely that pose is followed. For both sliders, higher values follow the reference more closely but stick less to your prompt.

Custom avatars#

In addition to the predefined avatars, uploading a custom avatar lets you upload a base image (that you own the rights to) for your avatar that will be used for all selfies. You can set it up from your Kindroid's settings under Avatar > Edit avatar. Your own user avatar, used in group selfies and couple autoselfies, is set on your active persona in Profile > Personas.

Generated avatars#

If you don't have a picture to upload, you can generate one. Open your Kin's settings, go to Avatar > Edit avatar and tap the grid icon at the top right to open Global generated avatars. Tap +, pick an Avatar Style (Photoreal, the default, or Anime), describe the look you want under Prompt (up to 200 characters) and tap Generate avatar. You'll go back to the gallery, and the new avatar shows up there once it's ready. When you create a new Kin, you'll also see a Generate avatar button while picking their look, which opens the same gallery.

Tap any avatar in the gallery to Use it, Copy prompt or Delete it. Use opens Configure Avatar with the image loaded, your prompt filled in as the avatar description and the matching style picked. You can add more detail to the description (see Avatar description below), then tap Set as avatar.

Generated avatars belong to your account rather than to one Kin, so all your Kins share the same Global generated avatars gallery. Generating avatars doesn't use selfie credits, but there's a cap on how many you can make in a short time: 5 for subscribers and 2 for free users. The count resets every 20 minutes.

Display photo#

You can set a separate display photo for your Kin in their settings under Avatar > Display > Photo. This is optional and only changes how your Kin appears to you in chat, menus, thumbnails, and the call UI (including the displayed avatar and background). Selfies, image generation, group selfies, and sharing will still use your main avatar photo. This is useful if you want your Kin to display a different look in conversation and calls without affecting how they appear in generated images.

Avatar Slots#

Pick a Kindroid > tap their name at the top of the chat to open their settings > Avatar > View slots

Each Kindroid can hold up to six avatars at once. Rather than overwriting a single avatar every time you want a fresh look, you can store several and flip between them on the spot, which is handy for keeping a few distinct appearances or personas ready to go.

To set them up, open your Kindroid's settings, go to Avatar, and tap View slots. Tap an empty slot to create a new look, or use Save current to a slot to store the avatar you have now. Tap a saved slot to Edit, Use or Remove it. You can also switch slots with Avatar slot under Additional controls when requesting a photo selfie, which changes the active avatar everywhere. Your display photo and display animation stay the same when you switch.

Avatar description#

Whereas default avatars have a preset description that tells the AI what the avatar looks like, you get to write your own for custom avatars. When generating selfies, this information will be used in the selfies engine to help portray what your Kindroid should look like. In addition, during text chats your Kindroid will be able to recall these details.

Technical tip: avatar description for custom avatar is incredibly important - see Atelier Selfies Guide for more concrete tips specific to the latest selfies version.

Face detail enhance & face detail prompt (photoreal only)#

The face detail slider controls how much detail should be put into postprocessing your Kindroid's face. At 0%, there will be minimal additional processing and as a result the facial expressions as well as face details such as skin, freckles, will be minimal. Higher than 0 values can sometimes result in very detailed faces with more emotional expressiveness that also come with a higher variance in terms of facial structure and other details, so it's up to you on how much variance you want vs. how much detail you want to enhance. Artistic prompts or prompts with more nuanced expressions do well with high face detail. Face detail prompt is additional directions for detailing the face. This should be where things like "green eyes", or "heavy freckles" are added, combined with a moderate to high face detail slider value. For more on this, see Tableau Selfies Guide

You'll find these under Advanced options in Edit avatar. The face detail slider is shown when your Kindroid uses Canon, Odyssey, Palette or Tableau, and the face detail prompt on Odyssey, Palette or Tableau. Neither appears when your Kindroid uses Atelier, the latest selfies model.

Avatar boost (Odyssey/V5 only)#

Avatar boost lets you train an adapter using at least 4 images that enhance consistency and likeness to the images you used to train with. The images you select are the most important part, and they should be high quality, and focused on what you want to emphasize. Images of an avatar with face visible are great, and it should be a mix of different qualities with similarity in between. Too homogenous of images might result in all images looking the same way - for example, having the same outfit in all photos might make your Kindroid always wear that outfit. We recommend keeping the face & hair and other immutable properties in most photos but switching up on other qualities. A mixture of portrait and zoomed out body shots at mostly frontal to three-quarter angles can work well for a general avatar consistency boost; too many images that are very closeup may not result in great results. Training an avatar boost takes a while, but after it's completed, you can apply a weight to it. 0.7 is default, and the right weight depends on how much consistency vs. freedom tradeoff you want to have. Higher weights may result in more limb mutations as a known side effect. Higher weights in groups may also result in more severe bleeding into other Kindroids (group selfies will also lower your set weights slightly for 2 and 3 person selfies to cause less bleeding between). Uploading more images will not train for more time - uploading more than 4 can help reduce too much emphasis being placed in particular traits of the 4 images and can smooth out the traits of images more, but 4 well selected images are sufficient for a great avatar boost.

Inpainting#

For a selfie, if you want to fix little things like wrong hands, or if you want to even change the outfit in an image, open the selfie and tap Actions > Edit Selfie. You can then paint over the area you want to change, describe it and tap Request Change (1 credit), which will result in a new selfie.

To properly use inpaint, you should describe what's in the picture, especially what should be in the selected area. Describing the holistic picture isn't necessary but can help with cohesion, but instead of saying "fix the mutated hand", write "proper, correct hands". Always frame things positively instead of what should not be there.

Autoselfies#

In text chat, tap your Kindroid's latest message and choose Autoselfie (subscriber only) to let your Kindroid come up with a prompt based on conversation context. You can help the context be a little more accurate by recapping for your Kindroid right before requesting the autoselfie. Pick Solo for just your Kindroid, or Couple to put your own user avatar in the shot with them (this needs an avatar with an Appearance description on your active persona in Profile > Personas). You can also get Kindroid-initiated autoselfies by turning on Proactive selfies in your Kindroid's settings. They will show up in chat by default, but can be turned off with Show in chat under Additional controls in the autoselfie menu.

Groupchats also have group autoselfies, where you can select up to 3 participants (including yourself) in order. They must be of the same avatar type (photoreal or anime), and works similarly to single autoselfies. Tap the last AI message and choose Group Autoselfie to open the group autoselfie menu. You can also select only 1 AI for an individual selfie but with group settings and context, particularly useful for scenarios.

Memory-aware autoselfies#

Autoselfies can now also draw on recurring visual details from your Kin's learned context, long term memories, journals, and key memories, things like a meaningful object, pet, decor, attire, or place. This helps them feel more grounded in your Kin's established world. Current settings and recent conversation still take priority, and since this depends on model behaviour, results will vary.

Animated Avatars#

You can animate both predefined as well as your own custom avatars. Once animations are set, you can click on the portrait to toggle the animation on/off state.

Predefined Avatars: You do not need to do anything here as they are already animated. Each come with its preset animations attached.

Custom Avatars: To animate your Kin, open their settings, go to Avatar > Display and tap Animation, then tap + to upload a photo and select Animate for 4 credits. Animating is a subscriber feature and can take some time to complete. Once it's ready, it shows up in your Animations library, where you can tap it and choose Use to show it on your Kin's portrait in chat and in calls.

Animations for custom avatars are kept separate from the avatar image itself, so the animation you see every day can be different from the image used as the reference for selfies. You can switch, unset, or delete animations from the Animations library.

Technical Note: Custom animations are still new and most avatars with human-like faces will work well, but stylized/anime/non-photoreal faces may have higher chances of undesired effects.

Image/video seeds#

For more advanced users, you may want to copy a "seed" from an image. This is available for selfies generated on Odyssey or newer engine only. You can copy a seed from an open selfie with Actions > Copy seed, and paste it into the Seed field under Additional controls when requesting a new selfie. Edits made with Edit Selfie also take an optional seed under More Controls.

A seed is the starting signature for an image, and using the same prompt and settings along with the same seed will result in the exact same image. This is helpful if you want to keep most of the composition but change a part of your prompt for minor variations in looks.

Technical note: when rerunning a selfie with Actions > Rerun Selfie, it will not reuse the seed. This is because if the exact same prompt and exact same seed is used, you'll just get the same image again.

Organizing and saving your gallery#

Open any selfie or video in your Kin's gallery and tap Actions to find these tools:

• Add Tags: add one or more tags, or type a name under Create a new tag to make a new one. To rename a tag, change its color or delete it, tap the ... next to it. Each Kin can have up to 100 tags.
• Save to device: on web this downloads the file, and on iOS it opens the share sheet so you can choose where to save it. On Android the option is called Save to Files and saves the file straight to your device.
• Flag Selfie (Flag Video for videos): Android only. Use it to report a result to our team. Enter a Reason for report of at least 12 characters, then tap Submit report.

To find things later, use the Tags filter at the top of the gallery. On phones and smaller screens it's in the More options (...) menu instead. Pick up to 10 tags to see everything that has any of them, choose Untagged to see only what you haven't tagged yet, or choose All tags to clear the filter.

To tag several items at once, open More options and choose Bulk select. Select the items you want, then tap Add Tags or Remove Tags. You can also Share the selected items to your profile in one post (up to 9, with only one video) or Delete them.

Filter for selfies on apps & differences with web version#

There are slight differences in selfies when requesting via the app vs. the website at kindroid.ai. There is a keyword-based NSFW filter on words that may trigger NSFW images on the app, but not when Kindroid is accessed from the website. The filter will be more robust for video selfies due to the existence of image input on apps. This applies to both prompts you write and to autoselfies prompts. This is a precaution for certain app store policies dictated by Apple and Google. Even if you subscribed on the app, you will have full access to the web version and vice versa, and aside from the differences covered in this section, the app and the web versions are functionally the same.

The filter does not exist on the web version of Kindroid. On the web version, you will also have the option to use the NSFW engine, which may produce better prompt adherence at some expense of visual quality.

The apps also check each result after it's made. This covers photo selfies, edits made with Edit Selfie, group selfies, autoselfies and video selfies that you request on iOS or Android. If the check finds nudity, the result is removed and you'll see Nudity detected - removed per App Store policies (Play Store on Android) in its place, and the credit you spent is refunded. Because the file is removed, it won't show up on web either.

The check can sometimes flag a result by mistake. The avatar description is a common cause, so look over it and your prompt before you try again.

Avatars you generate on iOS or Android are checked the same way. A flagged one shows as Blocked in Global generated avatars and can't be used.

Proactive selfies work a little differently. If you last messaged your Kin from the app, their proactive selfies get the same check, but a flagged one is locked instead of removed. In the app it shows as Nudity detected, while on the web it shows normally. To see it in the app too, open Kindroid on the web and tap Unlock for app on the selfie in chat, or open it in the gallery and choose Actions > Unlock for app. Unlocking only works on the web, and only for proactive selfies.

Understanding earned selfie credits#

Credit accumulation and limits:

• Paid Subscribers: 1 selfie credit every 30 minutes, up to 25 standard credits
  • Ultra and MAX add-on users will get 2 selfie credits every 30 minutes
• Free preview (new accounts without a subscription): 1 selfie credit to start, then 1 every 2 hours (holding up to 2 at a time) until 5 preview selfies have been granted in total

Daily rewards:

On top of the regular refills, subscribers also receive additional credits from daily rewards (Profile > Rewards), which reset at midnight PT every day.

Queue delay#

Photo and video selfies share one queue delay. It builds up in tiers when you make a lot of selfies in a short time:

• Each photo selfie, group selfie, autoselfie or edit made with standard credits adds 1 tier. If it uses paid credits, it adds none.
• Each video selfie adds 2 tiers with standard credits, or 1 with paid credits.
• One tier recovers every 5 minutes.

Once you reach your plan's threshold (2 tiers, or 10 with MAX), new requests are pushed back 5 minutes in the queue, plus 5 more minutes for each tier above it. This lets other people go first. The request screen then shows Queue delay: N minutes. Selfies made with paid credits still wait out any delay you've already built up, but autoselfies skip it. If no other requests are ahead of yours, yours can start sooner. Spreading your requests out helps you avoid the delay.

With MAX, your selfies get priority on dedicated compute, with little or no queue on the latest selfie model, until you've made about 10 in a short time. While priority is on, the request screen shows Priority selfie is enabled. After that, the standard queue delay applies.