Canadian AI image startup Ideogram, founded last year by former AI researchers from Google Brain, has made a new for itself among AI creators with its text-to-image models that produce a wide range of styles from realistic to fantastical, and most impressively of all, highly accurate text baked into the image itself (something other leading image generators, including Midjourney, took a while to implement and still struggle to generate reliably).

Now, it’s getting in on the trend of expanding its web-based user workspace to include a new interactive, infinite Canvas where users can spread newly generated images out, compare them to older generations, resize and reorder them at will, and even combine multiple AI generated images into one new composite.

It also allows users to upload their own visuals. With this addition, Ideogram Canvas aims to streamline workflows and offer flexible tools to refine creative projects step-by-step.

But of course, Ideogram is far from the only AI company to go beyond the simple chatbot-style text entry interface.

Earlier this month, OpenAI launched an experimental new “Canvas” view for ChatGPT. Unlike Ideogram’s version, it doesn’t help with imagery. Rather, OpenAI’s version offers the ability to see text-based documents and code alongside the chat interface, and watch as their chat conversation changes the resulting output in the “Canvas” view to the right.

Moreover, Ideogram’s “Canvas” view is closely reminiscent of an approach that was pioneered last year by a startup called Visual Electric, which uses open source Stable Diffusion AI image generation models, and which recently launched a mobile app. However, Ideogram trains and offers its own proprietary, ground-up image generation models such as the recently launched Ideogram 2.0, which sets it apart.

Magic fill and Extend

Alongside Canvas, Ideogram is also debuting two additional new features: Magic Fill and Extend.

• Magic Fill allows users to edit specific regions of an image by replacing objects, adding text, changing backgrounds, or fixing imperfections. The tool enables users to focus on particular areas of an image and generate high-resolution details with a simple text prompt.
• Extend helps users expand images beyond their original borders, keeping a consistent style. This tool is useful for resizing images, adjusting composition, or adapting content to different screen formats without losing the original structure.

These tools, designed to complement each other, give users the ability to make extensive edits or modifications to images while maintaining the overall quality and coherence of the content.

Subscription plans and features

Ideogram Canvas is available with all of Ideogram’s various usage tiers, though naturally paid plans get you more perks and features and fewer limitations. In fact, the company posted a thread on its social account on the network X (formerly Twitter) that noted all paid plans receive unlimited Canvases. The pricing for the various options is as follows:

• Free Plan: Allows up to 40 images per day with 10 slow credits, access to 2 canvases, and basic features like text-to-image generation and compressed image downloads.
• Basic Plan ($7 USD/month, billed annually): Offers 400 priority credits per month, 100 slow credits per day, unlimited canvases, and access to Magic Fill and Extend, along with features such as PNG downloads and customizable aspect ratios.
• Plus Plan ($16 USD/month, billed annually): Adds 1,000 priority credits, unlimited slow credits, image uploads, private generation, and additional customization options.
• Pro Plan ($48 USD/month, billed annually): Includes 3,000 priority credits per month, support for up to 12,000 images, and an upcoming bulk generation feature with CSV integration.

Furthermore, Ideogram offers its own API that developers can use to build third-party apps atop, yet this offers only the new Magic Fill and Extend features rather than the Canvas (which makes sense, since it is highly integrated into and dependent upon Ideogram’s website design). Pricing for accessing the models through the API ranges from $0.01 per input to simply describe images to $0.08 per input for image generations with Ideogram 2.

Ideogram credits part of the development of Ideogram Canvas to its community of beta testers and members of the Ideogram Creators Club, who provided feedback during the platform’s testing phase. The company acknowledges their contributions in refining the platform’s functionality and design.

Expanding its teams

As part of its broader growth strategy, Ideogram also noted it is expanding its teams and open to hires in Toronto and New York City.

The company is actively recruiting for various roles across AI research, engineering, marketing, and finance to continue developing its suite of AI tools. Interested candidates can apply via the company’s jobs page.

With the launch of Ideogram Canvas, the company seeks to offer a platform that blends user-generated content with AI-assisted tools like Magic Fill and Extend. By making it easier to create and modify images, Ideogram aims to support creators in a wide variety of industries.

The fundraising landscape is shifting fast, and in 2025, the old rules no longer apply. Startups navigating flat, down, or extension rounds must be more strategic than ever in securing their next round of capital. Traditional metrics like $100 million in revenue for an IPO or $1 million ARR for a Series A are becoming outdated. At TechCrunch Disrupt 2024, we’ll explore the evolving dynamics founders must understand to succeed in today’s challenging capital markets.

On the Builders Stage, we’ll be joined by an esteemed panel of experts who have lived through these changes and come out on top. Nikhil Basu Trivedi, co-founder and general partner of Footwork; Dayna Grayson, co-founder and general partner of Construct Capital; and Elliott Robinson, partner at Bessemer Venture Partners will share their insights on how startups can thrive amid unpredictable fundraising environments.

Meet the speakers

Nikhil Basu Trivedi, Co-Founder and General Partner of Footwork

Nikhil Basu Trivedi has built a reputation as one of the sharpest minds in early-stage consumer tech. At Footwork, Nikhil focuses on startups at the intersection of consumer behavior and enterprise technology, sectors that are transforming the way investors evaluate opportunities. His investments in Canva, Lattice, and other category-defining companies demonstrate his knack for identifying breakout potential even in difficult markets. At Disrupt, Nikhil will discuss how founders can differentiate themselves when traditional benchmarks don’t apply.

Dayna Grayson, Co-Founder and General Partner of Construct Capital

Dayna Grayson’s expertise lies in industries that have long resisted change, such as manufacturing and mobility. As co-founder of Construct Capital, she’s led investments in companies that are reimagining legacy sectors and securing funding in environments where growth is slow and investor confidence can be shaky. With experience guiding startups like Desktop Metal and Formlabs through various funding stages, Dayna will offer valuable insights into how companies in traditional industries can attract the capital they need to drive transformation.

Elliott Robinson, Partner at Bessemer Venture Partners

Elliott Robinson is a leading authority on cloud software, and his work at Bessemer Venture Partners has positioned him at the forefront of the SaaS industry. Co-author of Bessemer’s “10 Laws of Cloud Computing” and the annual State of the Cloud Report, Elliott understands the key drivers that investors look for in cloud and SaaS startups. From board seats at companies like Hinge Health and Canva to investments in Forter and Netlify, he has a bird’s-eye view of how cloud companies are navigating down rounds and market resets. At Disrupt, he’ll discuss how companies can secure funding, even during down rounds, by emphasizing the fundamentals that ensure long-term success.

Join the discussion on navigating today’s fundraising landscape

Whether you’re a founder navigating tough fundraising conditions or an investor looking for the next big opportunity, this conversation is one you can’t afford to miss. Join us at Disrupt 2024, alongside 10,000 tech, startup, and VC leaders, to gain actionable insights into the strategies that will define your next capital raise.

Don’t wait — tickets are selling fast. Be part of this essential dialogue at Disrupt 2024, happening October 28-30 at Moscone West in San Francisco. Get your pass before prices increase at the door. Grab the Expo+ 2-for-1 Pass, and your plus-one can join you for only half the cost of an Expo+ Pass. 

Microsoft and OpenAI announced they’re offering a select group of media outlets up to $10 million ($2.5 million in cash plus $2.5 million worth of “software and enterprise credits” from each) to try out AI tools in the newsroom.

The first round of funding will go to Newsday, The Minnesota Star Tribune, The Philadelphia Inquirer, Chicago Public Media, and The Seattle Times.

These outlets will receive a grant to hire a two-year fellow who will work to develop and implement AI tools using Microsoft Azure and OpenAI credits. The program is part of a collaboration between Microsoft, OpenAI, and the Lenfest Institute for Journalism, which aims to promote local media.

“While nothing will replace the central role of reporters, we believe that AI technology can help in the research, investigation, distribution, and monetization of important journalism,” Tom Rubin, the chief of intellectual property and content at OpenAI, said in the press release.

Microsoft and OpenAI will provide grants to three more media organizations at a later date. Some AI applications that the outlets will explore include leveraging the technology for, transcription, content summaries, and creating a “conversational” search tool for archives.

Google is continuing to infuse the Gemini Live AI assistant across Android phones with the ability to make calls and send messages, even when your phone is locked, on its way. As discovered in a piece of as-yet unreleased code by Android Authority, the update would further cement Gemini as the replacement for Google Assistant.

Controlling the phone while it is locked is a signature feature of Gemini Live. You could already talk to the AI and ask it questions when your phone is locked, but currently, only Google Assistant can send messages or make calls without unlocking the device. The unreleased code points to Gemini getting that option, too.

An AI system developed by Meta can find solutions to maths problems that have eluded mathematicians for over a century, researchers at the firm claim.

The problems involve mathematical tools called Lyapunov functions, named after mathematician Aleksandr Lyapunov, which analyse whether a system will remain stable over time, meaning its behaviour can be predicted. One famous example of such a system is the motion of three celestial bodies as a result of their mutual gravitational interactions – describing the behaviour of this “three-body problem” is extremely challenging.

The second season of the Legend of Legends animated Netflix show Arcane comes out next month, and Riot Games plans to celebrate that in League of Legends itself. Namely, it will add Ambessa Medarda, a character introduced in the show, as a new champion as part of Patch 14.22.

You might recall that Ambessa is the matriarch of the Noxian Merdarda family who first appears partway through the first season of Arcane to advise her daughter Mel. After Jinx’s attack on Mel and the rest of the Piltover council at the end of the first season, it seems like Ambessa will be on a path of vengeance in the show’s second season. Riot Games believes that she’s a compelling enough character to bring over to the game that inspired Arcane, with game designer Max Perlman saying that the show gave the developers a “jumping off point to eventually find the center point of her kit,” in comments provided to Digital Trends.

Ambessa is designed to be a top lane champion who can 1v1 against other characters, and her abilities reflect that. Her passive ability is Drakehound’s Step, which allows Ambessa to dash after casting an ability and give her next ability range, damage, and attack speed buffs while refunding energy. Her Q ability starts as Cunning Sweep, a basic attack that deals damage in front of her with blades; if Ambessa hits an enemy, it turns into Sundering Slam, which deals bonus damage to the first enemy it hits while doing damage in a line.

Ambessa’s W ability is Repudiation, which gives her Shield as she slams into the ground and damages nearby enemies. Her E ability is Lacerate, which deals damage around Ambessa and enables a combo with Drakehound’s Step by allowing her to strike twice. Finally, her Ultimate Ability is Public Execution, which teleports her to the farthest enemy champion in a line players choose and suppresses, damages, and stuns that champion.

Ambessa seems like an appealing choice for more aggressive players and newer ones playing the game for the first time after watching Arcane. Ambessa’s addition isn’t the only Arcane-related tie-in with League of Legends, as its All Random, All Mid game mode is getting a new map based on the Bridge of Progress and more between the launch of Patch 14.22 and January 8, 2025.

The second season of Arcane will be released across three acts on November 9, November 16, and November 23.