Ideogram is one of the flagships of the track, founded by the former Google Imagen team. Known for "the strongest text rendering", Ideogram 3.0 is extremely friendly to designers in terms of layout, posters, brand materials and consistent characters.

Ideogram

Core parameters and statistics of Ideogram

Ideogram was founded by a former Google Imagen core researcher. When the product was launched, it focused on the "strongest text rendering" segment. It took a completely different path from the artistic fantasy of Midjourney and the ecological binding of Adobe Firefly. Its parameter list looks similar to a traditional image generator, but each design serves the core scenario of "text + design".

Dimensions	Key facts
Product positioning	The most powerful image generator for text rendering + commercial design scene guidance
Model generation	Ideogram 3.0, 2.0, 1.0, Beta
Core Competencies	Text Rendering, Style References, Magic Prompt, Magic Fill/Extend/Replace, Character References
Output specifications	Multi-ratio (1:1/16:9/9:16/4:5, etc.), up to 1536×1536 output
Business Tier	Free → Plus($8/mo) → Pro($20/mo) → Studio($48/mo) → Enterprise/API
Covered platforms	Web, iOS, Android, API
Company endorsement	Led by Andreessen Horowitz, founded by former Google Imagen team

Positioning boundaries: Ideogram is not used to replace Midjourney for artistic exploration, nor is it used to replace Photoshop for refined synthesis. Its core battlefield is "visual content that requires clear text" - poster logos, banners, brand materials, social media cards, and packaging design. This type of scene has always been a blind spot in traditional AI image tools, and the text generated by most models is either blurred or garbled.

Ability Distribution: Ideogram's ability tree takes Text Rendering as the root node and grows upwards to Style References (brand style unification), Character References (character consistency), Magic Fill/Extend/Replace (design closure with editing) and Canvas (multi-image splicing). These capabilities are not superimposed in isolation, but are strung into a chain around the "brand design workflow" - from drawing to editing to puzzle to reuse, every link of the chain requires that the text is correct.

Benchmarking pattern: In the multi-polar pattern of AI image generation, Ideogram and Midjourney (art exploration), Adobe Firefly (design ecology), Stable Diffusion(open source community) forms a complementary rather than a substitute relationship. The moat of Ideogram is not the number of model parameters or artistic styles, but the cognitive barrier established on the single indicator "text rendering accuracy" - the designer group has equated Ideogram with "the most stable AI for making posters with text".

Users and market recognition of Ideogram

Ideogram's market recognition does not come from overwhelming advertising, but from the natural fermentation of the "Text Rendering Comparison Contest" triggered by each release in the X (Twitter) design circle. The designer group is the most demanding early user group, and their tolerance for typography errors is almost zero - Ideogram's reputation among this group is itself the strongest vote of confidence.

Community and word-of-mouth: During the public beta of Ideogram 1.0, the label "image generator that can write" received tens of thousands of reposts on X, and the design community spontaneously organized multiple rounds of text rendering blind tests of Midjourney vs Ideogram. By the time Ideogram 3.0 was released, many leading design media (such as UX Collective, Design Week) rated it as "the first choice of AI for brand designers".

Enterprise adoption: As of the current point of time, Ideogram has not disclosed the list of corporate customers and the specific number of contracts (not officially disclosed), but the following signs can be observed from the API ecosystem: multiple SaaS platforms (social media management tools, e-commerce design plug-ins, marketing automation platforms) have integrated the Ideogram API as the backend for text and image generation; some FMCG brands have introduced Ideogram into their internal material workflows to generate event posters and packaging renderings. B-side adoption is still in the early stages, which is directly related to the short official launch time of the API (August 2024).

Capital Approval: The company has received investment from top VCs such as Andreessen Horowitz. The founding team all comes from the core R&D team of Google Imagen (Mohammad Norouzi, Chitwan Saharia, Jason Baldridge, William Chan). This technical background is a strong credit endorsement when financing.

Competition Test: After the release of Ideogram, Midjourney has significantly enhanced its text rendering capabilities in its V6/V7 version, and DALL·E 3 has also continued to improve its typesetting performance. This shows that the success of Ideogram has forced the entire track to pay attention to the previously ignored demand of "text + image". However, in the mainstream horizontal evaluation so far, Ideogram still ranks first in English text accuracy.

Cost advantage of Ideogram

Ideogram's cost structure adopts a three-tier model of "free points + tiered subscription + API billing", with different payment paths designed for individual designers, studios and enterprises. The key to cost interpretation is not the absolute price, but "in scenarios where text rendering is required, Ideogram is the most cost-effective alternative to labor costs."

C-side/Personal Cost: The Free plan gives away basic points every day, which is suitable for light experience and evaluation. The Plus plan of $8/month is currently the most competitive level - at this price point, users receive commercial licensing, private generation and a monthly points pool. Compared with Midjourney the lowest paid tier (starting at $10/month) and Adobe Firefly's generation point system, Ideogram's text rendering capabilities plus commercial licensing make it a "closed-eyed" cost-effectiveness for independent designers.

Developer/API Cost: Ideogram API adopts a standardized billing model based on the amount of calls, and the pricing is not fully disclosed on the official website (requires registration or business consultation). Compared with similar image generation APIs (such as DALL·E 3 API, Stability AI API), Ideogram has a higher effective output rate in a single call in text-intensive scenarios - because high text rendering accuracy means low rerun rate. Hidden costs include: API concurrency limits, response time fluctuations (potential 3-5 second delays during peak times), and the scope of multilingual support for text content in output images.

Enterprise/Privatization Cost: The Enterprise plan requires business negotiation confirmation, involving terms such as SLA levels, private deployment possibilities, audit logs, batch management backend, etc. For companies that produce a large amount of brand materials (more than thousands of images per month), Ideogram's Studio plan ($48/month) can first be used as a cost anchor for mass production. However, when evaluating enterprise-level implementation, it is necessary to focus on confirming three hidden cost items: the throughput stability of the API during peak periods, the actual accuracy of text rendering for non-English languages ​​(CJK, etc.), and the copyright boundary of the output content in commercial usage scenarios.

Cost Perspective Summary: Ideogram's cost advantage is not reflected in "the lowest absolute value", but in "in terms of text rendering, which is a necessary capability point, it is currently the only option that does not require repeated rework." The time savings from reduced rework often make more financial sense than the subscription fee itself.

Main functions of Ideogram

Ideogram's functional system is designed around "brand design workflow" rather than "general image generation". If Midjourney is compared to an artist's workbench, Ideogram is more like a brand designer's pre-press output system - each function solves a real design delivery pain point.

• Text Rendering: Core moat capability. Ideogram can generate clear, accurate, and well-formatted English text in a picture, with controllable font style, weight, and alignment. In scenarios such as brand posters, logo proposals, and packaging renderings, the text no longer requires post-PS repair, and the final draft can be produced directly with a prompt. English currently performs best, but support for Chinese/Japanese and other CJK characters is still limited, and there is a risk of glyph aliasing when multi-language mixed typesetting.

• Style References: Upload 1 to 3 reference pictures, and Ideogram will extract their color, light and shadow, material, and composition style as generation constraints. This is crucial for brand visual consistency - the same set of brand colors, font styles, and visual language can be stably reproduced in posters, banners, and social media cards with different images, without the need to re-describe the style for each picture. Efficiency Quantitative Deduction: The brand design team switched from "manually writing style prompts for each picture" to "creating a Style Reference library for reuse". The prompt debugging time before a single picture is released can be reduced from 15-20 minutes to 2-3 minutes.

• Character References: After uploading character reference pictures, Ideogram can keep the same character's face and clothing consistent in different scenes, compositions, and actions. Suitable for virtual spokesperson IP characters, series illustrations and other scenes that require character reproduction. The current version performs better on the front and half sides, but there is still a risk of drift when extreme angles and expression changes occur.

• Magic Fill/Extend/Replace (local editing toolset): Magic Fill does local redraw modifications, Extend does image expansion to open a new composition, and Replace does element replacement. These three tools form a complete design pipeline from "generate → find defects → local repair → composition level editing", and users do not need to switch to Photoshop to modify pictures. Implementation Tips: In batch poster production, Magic Fill's "change the text without changing the background" ability is the key to saving rework time - copywriting modification changes from "re-running the entire image" to "partial replacement", and the single modification time is reduced from 3-5 minutes to 10-20 seconds.

• Magic Prompt (prompt word optimization): After entering a simple description, Ideogram will automatically expand into a more detailed and clearer structure prompt, improving the stability and consistency of the image quality. It is especially friendly to non-native English users or prompt novices, reducing the hidden cost of "trying prompt".

• Canvas (canvas splicing): Drag multiple generated results to the same canvas for composition and combination. Suitable for moodboards, comparison displays of multiple solutions, and combining multiple design elements into a complete picture. Used in conjunction with Style References, it can quickly produce multi-scenario presentation boards for brand visual proposals.

• Image Remix: Adjust style, text content or composition based on existing generated results instead of regenerating from scratch. It is suitable for comparison and acceptance of the "fine-tuned version" and the "original version" in design iterations.

Functional Synergy: The above functions are not a stack of tools used in isolation. A typical design workflow is: Use Style References to lock the brand visual baseline → Magic Prompt to quickly produce a first draft → Magic Fill to modify copy or elements → Canvas to splice multiple solutions → Character References to do series extensions. This process compresses "brand design material production" from a serial manual process into a parallel AI-assisted link.

Ideogram model and version evolution

The evolution of Ideogram's versions is clear. Each major version corresponds to a significant jump in text rendering capabilities. At the same time, it also gradually completes the puzzle of abilities from "being able to write" to "being able to design".

Public beta period: verify market demand

Ideogram Public Beta (2023-08-22) is the first time the project is open to the public. The selling point at that time was very focused - "an image generator that can write". During the window period when neither DALL·E 3 nor Midjourney V6 had perfected text rendering, Ideogram quickly became popular in the X design circle with this single capability, verifying that the need for "clear text rendering" does exist and has not been met.

Mainline Release: Ability Ladder

• Ideogram 1.0 (2024-02-28): For the first time, English text rendering is stable and available. Designers began to incorporate it into real workflows - previously, AI-generated pictures with text basically relied on post-PS handwritten text. Ideogram 1.0 compressed this step from "1 hour P picture" to "30 seconds to generate".
• Ideogram 2.0 (2024-08-21): The realism and composition quality are greatly improved, and the text stability is further improved; the official API and iOS/Android mobile terminal are also released. The release of this version marks the evolution of Ideogram from "being able to write" to "being able to draw realistic pictures" and opening up the commercial API path.
• Ideogram 3.0 (2025-03-26): The latest official version. It has been fully upgraded in four dimensions: realism, composition, text rendering, and style consistency. New editing tools such as Style References, Magic Fill / Extend / Replace have been added, making the product evolve from a "simple image generator" to a "brand design tool with editing capabilities". Version 3.0 further consolidates Ideogram’s leading position in mainstream text rendering evaluations.

Version observation

The version rhythm of Ideogram (about 6-7 months for one major version) is slower than that of Midjourney (about 3-4 months), but the changes in each major version are larger. This shows that the team is doing in-depth research and development on each capability line rather than quickly stacking functions. The evolution path from public beta to 3.0 shows a clear strategic intention: first do transparent text rendering to establish barriers, then do realism and composition to level the gap with competing products, and finally use editing tools and brand functions to build workflow closures.

Technical advantages of Ideogram

Ideogram's technical advantages cannot simply be summarized as "the strongest text rendering", but need to understand "why it is the strongest text rendering" and "what does this mean to designers".

Mechanism - Differentiated architecture for text rendering: Text rendering in image generation is a problem that is difficult to be perfectly solved by a typical "end-to-end" model - when the diffusion model generates images, the text area needs to meet the three constraints of "accurate glyphs", "reasonable position" and "consistent style" at the same time. The text-image alignment technology accumulated by the Ideogram team during the Imagen period is the foundation, but more importantly, they have added an attention enhancement mechanism dedicated to text areas into the model architecture, allowing the model to obtain more character-level supervision signals when generating text areas. This is fundamentally different from the way the general diffusion model treats text as part of the image texture.

Effect - from "can write" to "can be typed": The text rendering effect of Ideogram 3.0 has surpassed the stage of "readable text" and reached the level of "available for typesetting" - the word spacing, line height and alignment of the text basically meet the design specifications under regular poster sizes. This means that designers no longer need to make secondary PS corrections for the text area of ​​each picture, thus saving 30-50% of the post-processing time for a single picture.

Technical implementation significance of Style References: Style References is not just as simple as "uploading reference images and then imitating the style". What it does at the technical level is "cross-image style decoupling and reorganization" - the model extracts style factors such as color distribution, light and shadow patterns, and material characteristics from the reference image, and then independently combines these factors with the semantic information in the generated prompt. The actual effect of this is: after the user uploads a brand visual specification map, all subsequent outputs can automatically align with the brand visual baseline, and there is no need to repeatedly describe "Please use the red XX font XX light and shadow of the XX brand" in the prompt.

Cycle of editing tools has value: Magic Fill/Extend/Replace is not a collection of independent functions. They form a design feedback loop of "generate → verify → modify → confirm". This cycle reduces the pressure of "getting it right once and for all" at the technical level - designers no longer need to rerun the entire image just for a single word mistake, but can accurately fix it during the editing stage. This improvement in workflow efficiency is multiplicative.

Technical boundaries for implementation: The above technical advantages are best in the English scenario. The rendering accuracy of CJK text (especially Chinese and Japanese) is still significantly lower than that of English, and glyph conflicts or alignment problems may occur when mixed typesetting in multiple languages. This is a difficulty that has not been completely solved at the technical architecture level - the stroke complexity of Chinese characters is much higher than that of English letters, and the resolution requirements for the attention mechanism are also higher. The team has not disclosed the timetable for CJK optimization at this time.

How to use Ideogram

Ideogram provides four usage portals, covering all scenario needs from personal experience to enterprise integration. Each portal has different capabilities and cost structures.

Entrance	Adaptable Crowd	Core Competencies	Cost
Web (ideogram.ai)	All users	Full generation + editing capabilities, Canvas stitching	Free/Subscription
iOS / Android application	Mobile creator	Publish, save and share pictures on mobile	Free/Subscribe
API	Developer/Enterprise	Standardized commercial API, integrated into own products	Volume/Business
Third-party integrations	SaaS platform users	Indirect use through tools that have integrated Ideogram	Platform dependent

Typical usage process (brand designer scenario): Log in to the Web → Upload brand visual specification drawings (1-3 pictures) in Style References → Enter the design requirement text + enable Magic Prompt → Select the output proportion and style orientation → Generate a first draft → Use Magic Fill to adjust copywriting details or element positions → Use Canvas to combine multiple solutions for comparison and display → Export in batches after confirmation. In this link, Style References is the section with the heaviest initial investment (about 5-10 minutes to establish the brand style library), but the subsequent drawing time for each picture can be compressed to 30-60 seconds.

Typical usage process (API integration scenario): Register an Ideogram account → Obtain the API Key in the developer backend → Construct the request body according to the API document (including prompt, style_ref, size, quantity and other parameters) → Integrate into the image generation service on the product backend. You need to pay attention to the request frequency limit and response timeout settings of the API.

Quantitative deduction for cost reduction and efficiency improvement (human-computer collaboration): Taking the daily production of 50 social media posters by the brand design team as an example, the traditional process (designers manually write prompts, debug PS and repair text one by one) takes about 8-10 hours; using Ideogram Style References + Magic Prompt to produce the first draft and Magic Fill to correct details, the overall time can be compressed to 2-3 hours. This 60-75% time reduction mainly comes from "the text no longer needs to be repaired later" and "the brand style is set once and reused throughout the process". The efficiency gain here is based on functional logic deduction, and the actual gain depends on the team's maintenance quality of the Style Reference library and prompt engineering proficiency.

Human-machine collaboration boundary: The following section recommends retaining manual confirmation points - brand compliance review (text content, logo position, brand color accuracy) before final delivery, text content involving contract terms or legal statements, and key visual delivery to high-end customers. Ideogram can be responsible for "draft production" and "batch first drafts", but "final draft confirmation" and "original breakthrough" still require the aesthetic judgment and brand experience of human designers.

Product Pricing for Ideogram

Ideogram's pricing system follows a linear upgrade structure of "free experience → personal payment → studio → enterprise customization". Each level is gradually liberalized in terms of points amount, concurrency, management capabilities and business terms.

Free Tier (Free): Basic points are given daily and can be used for regular generation operations. It is suitable for the boundaries of personal experience capabilities and to evaluate whether the text rendering quality meets workflow requirements. The credit limit of the free tier means that it cannot be used for mass production, but it is enough to complete the test verification of 5-10 pictures.

Personal payment tier (Plus $8/month Pro $20/month): The Plus tier is the main choice for independent designers - $8/month includes a monthly points pool, private generation (generated images are not public), and commercial licensing. Pro greatly increases the amount of points, adds priority generation permissions and batch export functions. From a cost-effective perspective, Plus has covered most of the daily needs of freelance designers, while Pro is more suitable for heavy users who produce more than 20 pictures per day.

Studio Tier (Studio $48/month): Designed for small design studios and small teams. It includes a larger monthly points pool, brand asset management functions (shared Style Reference library by multiple people), and basic team collaboration capabilities. For a design team of 3-5 people, the diluted single-person cost of the Studio plan is about $10-16/person·month, which is about the same cost as the Plus plan of independent designers but with additional collaboration features.

Enterprise layer (Enterprise + API): Business consultation is required for confirmation. Involves SLA levels, private deployment possibilities, auditing and usage statistics, batch management backend, and possible custom model fine-tuning. API pricing is based on call volume, and the specific unit price is not displayed on the public page - you need to register a developer account or contact sales to obtain a quotation.

Purchasing Suggestion: First use the Free plan to complete 10-15 real design scenario tests (especially to evaluate the most frequent text rendering needs in your business). After confirming that Ideogram's text accuracy and style consistency meet the requirements, start the pilot with Plus or Studio. Enterprises need to verify three points with the official before purchasing: the throughput of the API during peak periods guarantees the actual accuracy data of CJK text, and the coverage of the copyright licensing terms of the output content.

Application scenarios of Ideogram

The core adaptation scenario of Ideogram is the production of all "visual content that requires clear text". The following three types of scenarios can best utilize its capabilities and advantages, and also have clear and verifiable benefits.

• Brand Marketing Posters and Social Media Cards: The marketing team needs to produce a large number of event posters, product promotion pictures, and social media cards every week. The synthesis of "pictures + copywriting" in the traditional process inherently requires repeated debugging. Ideogram's Text Rendering + Style References combination reduces the single drawing time for this type of task from 30-60 minutes to 3-5 minutes. Verification method: Select 5 social media posters from the past week, use Ideogram to regenerate them, and conduct an A/B test with the original version. Pay attention to the two indicators of click-through rate and completion time.

• Packaging and label design rendering: Packaging design needs to present the product name, selling point copy, ingredient description, and brand information at the same time in a limited space - the requirements for text rendering are extremely high. The advantage of Ideogram is that "one generation is close to the final draft". Designers can quickly produce multiple packaging solutions for customers to preview, instead of adjusting them word by word in PS. Note the boundaries: Legal information on packaging (list of ingredients, production date, etc.) must be reviewed manually. The accuracy of AI-generated content cannot yet completely replace manual proofreading in such scenarios.

• Extension of brand visual system: After the brand VI manual is determined, it needs to be extended to dozens or even hundreds of touch points (business card PPT template, official account cover, offline posters, display racks, industrial badges, etc.). Ideogram's Style References allows teams to build a brand style library and then batch generate first draft materials for each touch point, compressing the execution cycle of brand extension from 2-3 weeks to 3-5 days. Human-computer collaboration mode: AI is responsible for "mass production of the first draft", and the designer is responsible for "draft selection + refinement + final review". AI covers 70% of the execution workload, and humans cover 30% of decision-making and quality control.

• Scenarios where it is not appropriate to use Ideogram: artistic creation and abstract expression (Midjourney or DALL·E should be used), deep image synthesis and refinement required (Photoshop or Firefly should be used), designs where CJK text accounts for more than 50% (the Chinese rendering effect of the current version has not yet reached the English level), and scenes that require the generation of hyper-realistic portraits with extremely high requirements for identity consistency (the stability of Character References still has boundaries).

Applicable groups of Ideogram

The value of Ideogram varies greatly for users in different roles. The following are the three most suitable groups of people and the corresponding unfit boundaries.

• Graphic and Brand Designers: This is Ideogram’s core user group. Designers face a large number of "copywriting + graphics" integration needs every day, and Ideogram's text rendering capabilities directly hit the pain points. The value is reflected in: the drawing speed is increased by 3-5 times, copywriting modification no longer requires re-running the entire drawing, and brand style consistency can be systematically managed through Style References. Not suitable for borders: If the designer's main work is pure art illustration, hyper-realistic character photography, or high-end customized vision (each picture requires a unique creative concept rather than template production), Ideogram is not a better choice than Midjourney.

• Marketing and Growth Operations: Social media operations, growth teams, and event marketers need to frequently produce promotional materials with text. The advantage of Ideogram is that "one person + one tool" can complete the complete output from copywriting to visuals. There is no need to find a designer to produce pictures after the copywriting is written. Prerequisite: The team needs someone who can maintain the Style References brand library (at least the designer needs to be involved when it is first built), otherwise the output visual style may deviate from the brand specification.

• Small e-commerce and independent entrepreneurs: If the design of self-owned brand product images, event posters, and promotional cards are all outsourced, the cost per piece will range from 200 to 500 yuan. The subscription fee of the Ideogram Plus/Studio plan ($8-48/month) can cover the generation needs of dozens to hundreds of images, and the price/performance ratio is much higher than the outsourcing model. Hidden costs: Although the drawing speed is faster, screening, fine-tuning, and aesthetic control still require human investment-this process cannot be fully automated.

• Not applicable to the group: Users who need in-depth long text content with pictures and text in non-English languages ​​(especially CJK dense text), art directors who need highly customized original visuals and do not accept the "homogenization" of AI generation, and government and enterprise customers who require complete offline/intranet deployment and are not satisfied with the API calling model (whether the Enterprise plan supports privatization requires business confirmation).

Summary and Outlook

Ideogram's core competitiveness is based on the single barrier of "the strongest text rendering", and it has been gradually expanded into a "brand design workflow platform" through supporting functions such as Style References, Magic Fill/Extend/Replace, and Character References. It occupies a clear and stable position in the multi-polar pattern of AI image generation - it does not compete with Midjourney for art, nor does it compete with Adobe for ecology, but rather serves as the definer of the "text + design" subdivision track.

The current main limitations are reflected in three aspects: First, the rendering quality of non-English text such as CJK does not yet meet commercial standards, which limits its penetration speed in the Greater China, Japan and South Korea markets; second, the stability and precision of the editing tools (Magic Fill/Extend) are still far from Photoshop-level refinement, and manual secondary processing is still required in complex scenes; third, enterprise-level adoption is still in its early stages, and the API's large-scale cases and SLA guarantee system have not yet been fully verified.

Evolution directions worthy of attention in the future: further breakthroughs in realism and artistry after Ideogram 3.0 (narrowing the subjective aesthetic gap with Midjourney); the deep penetration of Style/Character References in corporate brand production lines; the depth of integration with design/e-commerce platforms such as Figma, Canva, and Shopify after the commercial scale of API; and whether the team will invest in special optimization of CJK text rendering - this will be Ideogram The key battle is whether it can truly become the "global designer's first choice".

Procurement/Adoption Risk Assessment: For teams planning to introduce Ideogram, it is recommended to proceed according to the rhythm of "pilot first and then expand" - select 1-2 high-frequency scenes (such as social media posters, event banners) and run them through the Free solution to verify whether the accuracy of text rendering meets business standards, and then upgrade Plus/Studio to enter mass production. Before enterprise-level procurement, it is important to confirm with the official: the throughput limit of the API during peak business periods, the current accuracy status and optimization roadmap of CJK text rendering, and the copyright ownership and responsibility definition of AI-generated content in the commercial use authorization terms - these three items are most likely to trigger compliance risks after implementation.

Comparison of competing products

Comparison dimensions	The tool	Competitor A	Competitor B
Core Differences	—	—	—
Price	—	—	—
Target User	—	—	--

Version Info

• Ideogram 3.0 ：Ideogram 3.0 has been fully upgraded in terms of realism, composition, text rendering, and style consistency; the new Style References allows users to upload reference images to fix the brand vision, and the editing tools such as Magic Fill / Extend / Replace are stronger. The model has become the benchmark for various "text rendering" evaluations. （2025-03-26）
• Ideogram 2.0 ：Ideogram 2.0 takes a significant lead in realism and text stability, and has released official API and mobile terminals. （2024-08-21）
• Ideogram 1.0 ：Ideogram 1.0 is released and open to the public, making "AI writing decent English text" a reality for the first time. （2024-02-28）
• Ideogram public beta ：Ideogram was released in public beta, positioning itself as an "image generator that can write" and quickly became popular in the X/design circle. （2023-08-22）